ANTHROPOLOGICAL RECORDS 5:2 aTREE CENTURIES OF WOMEN'S DRESS FASHIONS. A QUANTITATIVE ANALYSIS BY JANE RICHARDSON AND A. L. KROEBER UNIVERSITY OF CALIFORNIA PRESS BERKELEY AND LOS ANGELES 1940 REE CENTURIES OF WOMEN'S DRESS FASHIONS A QUANTITATIVE ANALYSIS BY JANE RICHARDSON AND A. L. KROEBER ANTHROPOLOGICAL RECORDS Vol. 5, No. 2 ANTHROPOLOGICAL RECORDS EDITORS: A. L. KROEBER, E. W GIFFORD, R. H. LowIE, R. L. OLSON Volume 5, No. 2, pp. I I -154, I I diagrams Transmitted April 12, 1939 Issued October 5, 1940 Price, 50 cents UNIVERSITY OF CALIFORNIA PRESS BERKELEY, CALIFORNIA CAMBRIDGE UNIVERSITY PRESS LONDON, ENGLAND The University of California publications dealing with anthro- pological subjects are now issued in two series. The series in American Archaeology and Ethnology, which was established in 1903, continues unchanged in format, but is restricted to papers in which the interpretative element outweighs the factual or which otherwise are of general interest. The new series, known as Anthropological Records, is issued in photolithography in a larger size. It consists of monographs which are documentary, of record nature, or devoted to the presentation primarily of new data. MANUFACTURED IN THE UNITED STATES OF AMERICA CONTENTS Page roblem ..................................... . 111 Te measures ..... . 112 Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 ?Match of old and new data ........... ... ... .... ... ... .. . 113 surement data ................................... . 117 'scriptive history of the proportions of dress ...... . . . . . ......... . 127 Width of skirt ....................... ............ 127 Length of skirt ..................... .... ........ 129 Position of the waist ..... 129 Diameter of the waist ............. .... ... ........... . 129 Decolletage ....................... .... . ....130 Width of decolletage ................ .... ... . . ......130 riodicity ...................................... 130 Skirt length .................................... . 130 Skirt width ...................................... 131 Waist length ..................................... 132 Waist width .................................... . 132 Decolletage .132 Decolletage width. 134 Comparison . . . . . . . . . . . . . . . . ... . . 134 cillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 errelations of dimensions ............. .... .... ..... ... . 136 iability and stability of style . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Year-to-year variability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Variability within the year .......... . . ............ ... . . 141 Causality of change ............ . .................... 147 inclusions ..................................... . 148 151 TABLES ability of observers ..... . . . . . . . . ...... . . . . . . . . . . . . . 113 ability of sources . 116 o of dress diameters to height of figure: year-by-year means, 1787-1936. 117 -by-year means, 16D5-1786 . . ..... . ........ . ...... . . *..... 120 year averages, 1605-1936 ..... . . . . . . . . . . . . . . . ........ . . . 122 -year averages, 1788-1936 .......... 123 year averages, 1788-1936, by decades ending in 6 . 124 -year moving average of dress diameters, 1788-1934 .... . . . . . . ...... . . 125 a of maximum and minimum skirt length .... . . . . . . . . . . . ....... . . . 131 of periodicity in skirt length .... . . . . . . . . . . . . ....... . . . 131 a of maximum and minimum skirt width .... . . . . . . . . . . ......... . . 132 of periodicity in skirt width .... . . . . . . . . . . . . . ....... . . . 132 odicity in waist length ... . * la.... . . ............ ............ 132 lodicity in waist width ..... . . . . . . . . . . . . . . . ......... . . . 133 Lodicity in decolletage ..... . . . . . . . . . . . . . . . . . . . . . . ..... 133 odicity in decolletage width .... . . . . . . . . . . . . . . ........ . . . 134 ison of six periodicities .. . . . . 134 relations of certain dimensions, by five-year periods, 1788-1936 .. . .... . . . . 136 odicity of dimension interrelations .... . . . . . . . . . . . ......... . . 136 entage deviations of actual year means from trend, 1788-1934 ... . . . ...... 137 -year averages of annual deviations from trend, 1788-1934 .... . . . . .... .. . 139 year averages of annual deviations from trend, 1788-1934 ... . . . . . .... .. . 139 ency of fluctuation units, 1788-1934 .... . . . . . . . . . . . . .... .. . . . 141 entage sigmas of annual means, 1787-1936 .... . . . .... . . .... . ... . . ..............a141 -year averages of percentage sigmas, 1787-1936 .... . . . . . . . . . ... . ...... 143 -ear averages of percentage sigmas, 1787-1936 .. * .. . . .............a... 143 ribution of size of variability coefficients among the si measures . .143 ma of five-year averages of coefficients of variability ................. . ....... 144 emes of dimension and variability by period ....................... . . . 147 [ ~~~~~~~~~~[iii] iv ANTHROPOLOGICAL RECORDS FIGURES 1. Vertical dimensions, 1787-1936, moving-average and year values . . . . . . . . . . . . 2. Horizontal dimensions, 1787-1936, moving-average and year values . . . . . . . . . . 3. Dimensions 2 and 3, by ten-year means, 1605-1936 . . . . . . . . . 4. Dimensions 4 and 5, by ten-year means, 1605-1936 . . . . . . . . . . . . . . . . . . 5. Dimensions 7 and 8, by ten-year mealis, 1605-1936 . . . . . . . . . . . . . 6. Skirt width and waist width, 1787-1936 ...................... 7. Frequency of deviations from trend, by fluctuation units, 1788-1934 8. Variability and amplitude of dimension, skirt, 1787-1936 . . . . . . . . . . . .. 9. Variability and amplitude of dimension, waist, 1787-1936 . . . . . . . . . . . . .. 10. Variability and amplitude of dimension, decolletage, 1787-1936 . . . . . . . . . . 11. Clustering of extremes related to variability, 1605-1936 . . . . . HREE CENTURIES OF WOMEN'S DRESS FASHIONS A QUANTITATIVE ANALYSIS BY JANE RICHARDSON AND A. L. KROEBER I. THE PROBLEM is an attempt to define stylistic data for a study of the problem of how stylistic an objective and quantitative man- or aesthetic changes prove to take place when dress fashions were chosen for in- they are examined quantitatively instead of rather than works of pure or in- through subjective intuition or feeling. It can- is due to the greater ease of not of course be asserted that the change be- erial which is not only fairly havior of women's evening dress would follow the strictly comparable from decade to same patterns as style changes in painting or n from century to century. This music or even in some other type of dress. But +nforces that the range of material any findings will presumably have some signifi- cannot compare landscapes with cance for the wider problem of how aesthetic pheric treatments with portraits. styles change in general; to which in turn we rial must not be too utilitarian: must have some answer before we can hope to in- ;period may be primarily objects quire fruitfully why they change. (degree of state, in another they The investigation had its beginning in a brief ely consider comfort or serve for article by Kroeber in 1919.2 The techniques of en's evening or formal dress has examination there developed--which will be ex- airly constant function for several plained in a moment--are now applied to a much t the same time it is about as free larger body of material. The assembling of this ian motivation as dress can well new material was the first contribution of re, for well over a century it has Richardson. Whereas the earlier article covered 'and published in fashion plates, the seventy-six years from 1844 to 1919, the en been preserved where most of present study carries on to 1936 as well as back 8themselves have long since perished to 1787, continuously except for two years (1822, c?essibly scattered. In short, 1833) for which no data were encountered. This as a topic for investigation, doubles the span for continuous data. Back of advantages of representing an art 1787, contemporary portraits and pictures had of the highest order is rela- largely to be substituted for pre-wear fashion self-sufficient; relatively plates, and they run fewer; but a fair set of ed or warped by considerations of specimens was assembled back to 1605. Our total lity; specific and uniform enough time range is thus three hundred and thirty-two ble from one period to another; of years--longer, we believe, than in the over- oh precludes complete repetitive whelming majority of statistical studies in eco- ion and stand-still; and on which, nomics. To be sure, the seriation is badly broken le industry of search, there can before 1787. The decade 1631-40 yielded twenty- ed fairly adequate information over one available illustrations, the double decade of time. 1691-1710 none at all; 1711-20, twenty, but 1721- these reasons that this type of 30 only three. Our pre-1787 findings are there- been chosen for study, rather than fore far less significant and reliable than those y special importance or interest since 1787. Our more detailed analysis is ac- possess in itself. In other words, cordingly based wholly on the last one hundred a convenient and promising set of and fifty years. But the findings made there, projected backward, and supported by the inter- ir kindness and generosity in offer- mittent materials over the preceding one hundred llections of original fashion plates and eighty-two years, allow some tentative con- terial for this survey, the authors clusions for the whole span of three hundred and ess their appreciation to the De- thirty-two years. Decorative Art, and the late Mr. .ter, of the University of California To the figures computed and plotted year by Howell, Mrs. Morton Gibbons, and the year, we have added a five-year moving average, p Company, of San Francisco; and to Erioson, of Berkeley. Assistance in 2 tion of these materials was furnished On the Principle of Order in Civilization Bonnel of Works Projects Administra- as Exemplified by Changes of Fashion, Amer. al Project No. 665-08-3-30, Unit A-15. Anthr. 21:235-263. Cited hereafter as Kroeber. F tlll~~~~~~~~~111] 112 ANTHROPOLOGICAL RECORDS which of course smoothes out the mere annual No. 2, length of skirt or dress. variations and gives a much more vivid picture No. 3, length of waist. of the trend of fashion at any one time. On the No. 4, length (or depth) of decollet other hand, the deviation or fluctuation of each No. 5, width of skirt. year's style from the average for five years is No. 7, width or thickness of waist. also much more clearly brought out by this new No. 8, width of decolletage. device. This annual fluctuation is obviously a measure of the stability of the style. Originally another measure was made: Another type of variability is that within maximum width of skirt if this width oce the year. How different are the several dresses above the hem. This was soon dropped as of one year from one another, as expressed by regular in occurrence, and it is mentio their "sigma" or standard deviation from their to account for the gap in the numbering mean? The sigmas as compared over a period of stylistically, as in recent years, this years express the changes in variability. may be of importance. In short, we have worked out quantities which In detail, the measurements were exec express the extremes of certain features of follows: women's dress style; the times of these extremes and the intervals between them; the rapidity and No. 1, or base: Total length of fi consistency of the trends of change; and the de- the center of the mouth to the tip of t gree of homogeneity or stability of the style toe. No. 2: Distance from the mouth to thel both in a given year and over longer spans. of the center front of the skirt. No. 3: Distance from the mouth to the diameter across the waist. The girdle, THE MEASURES lower edge of the corsage part of the coincide with this or lie above or belo The traits or features of dresses dealt with diameter. The girdle and edge have been number six. These comprise three vertical and garded because neither is a permanent C three horizontal diameters: of the skirt or No. 4: Depth or length of the decoll measured from the mouth to the middle o dress as a whole, of the constricted middle or per corsage edge in front. waist, and of the decolletage or cut-out at the No. 5: Diameter of the skirt at its neck. We are really examining the dimensions of base. the silhouette of the whole dress. There are No. 7: Minimum diameter in the region many other features of probably equal signifi- waist. See comment under No. 3. cance, and of which fashion is perhaps even more No. 8: Width of the decolletage acro conscious: trains, sleeves, girdles, flounces, shoulders. yokes, and so on. All these however come and go. They are never permanent, but sooner or Full-face or nearly full-face figure later disappear completely for a time. This used so far as possible. If the cases means that only short-range comparisons can be were few, profiles and near-profiles we instituted for them. The skirt and waist di- cluded. A side view eliminates decolle ameters, however, and in full dress the decol- breadth, and, if the forearm is held ho letage, cannot be escaped, as long as the funda- as is frequent in some periods, one or mental style of women's wear remains at all. It measures may also be lost. Otherwise, is this permanence of the six silhouette dimen- and especially semiprofiles, seem most sions that has led to our confining attention in fashion plates results not very difC to them. full-face views. All measurements were made on fashion plates In the seasons covered.by monthly fa or other pictures with calipers and ruler in journals, the winter months are of cour millimeters. To render them comparable, they ones in question for full dress. So f had to be reduced to a common standard. For sible, illustrations were sought in the this the "total length of figure" was chosen and to March issues. If these did not suCC recorded as measurement No. 1. The six dimen- April to June numbers of the journal we sions were then converted into percentage pro- amined, or preferably the December and portions of this. It is these percentages that issues of the preceding calendar year. are presented and dealt with throughout. It stance, January-March of 1850 plus Nov seems useless to publish the raw or absolute December of 1849 have been counted as 1 measures; but they have been preserved. Actu- An absolute requirement was that ea ally, the basic measurement No. 1 is not the lar figure be dated in a specific year. whole length of figure, but the length up to the structed "typical" fashions, even if Ca middle of the mouth. The top of the head does year, such as abound in most histories:: not answer, because of varying increments of were of course of no use. Moreover, nc coiffure and headdress. mate datings were included. The only- All six of the dimensions are maximum diame- occurs in the case of a few Van Dyck ters. They are as follows: portraits, whose dates are known to Cal RICHARDSON AND KROEBER: THREE CENTURIES OF .4OM1EN'S DRESS FASHIONS 113 a of several years. As the pre-1787 data and the Ankerman engravings for the Ladies' een used only in ten-year blocks, these National Magazine, of London. mations would not matter unless the Seasonal fashion plates are scarce, however, lapped over from one block into another. before the French Revolution, not only on ac- erlaps have not been used statistically, count of their age, but also because they were that a group of ten Watteaus dated 1710- not published to a great extent. Hence we be- been included in the 1711-20 average. came increasingly dependent on painters and en- old Kroeber measurements for 1844-1919 gravers, such as Winterhalter, Debucourt, Rey- imited to ten 'figures for each year. nolds, Moreau le Jeune, Chodowiecki, Boucher, ly some of these ten failed to show one Nattier, Fragonard, Hogarth, Watteau, Terborch, of the six diameters. Richardson sought Codde, Velasquez, Van Dyck, and minor painters examples as possible, so that there of the Dutch and French schools. not be less than ten cases for each meas- The idealized lithographed fashion plates from V. This was not a hard task for the last 1789 on, whether published in Paris, Karlsbad, and a half. Back of 1844, however, ten or Vienna, are strikingly uniform. There are per year became a rarity; ten full-face changes in face and pose only with the advent of badly ever occurred. Therefore it was the wood-cut and the zinc-engraved ink drawing to get perhaps six dimensions from one and photograph of recent years. The earlier content oneself with two from the next, painters are less subject to conventionalization b that in the end none of the dimensions than the lithographers. Faces and attitudes are e wholly unrepresented for the year. individualized, waists are thicker, and the ex- eless, there were richer years: 1799 cellent likenesses are often far closer to the twelve illustrations; 1809, fifteen; photograph of today than to the draftsman's or wenty-six. lithographer's formalized delineations. The change from lithograph or drawing to photograph is comparable to the reversed change from paint- SOURCES ing to lithograph. Thus we may say of our span of three hundred and thirty-two years, that the ill out the years from 1920 to 1936, data are rather conventionalized for the hundred d Harper's Bazaar were consulted, and, years 1789-1889, but are tempered with realism Costume Royal. Each of these American before and after. es devotes a great deal of space to the ns of Paris designers. If, however, re not sufficient of these Paris models ben year, the gap was filled by unsigned MATCH OF OLD AND NEW DATA style plates. 1844, it was necessary to go from one Although the measurements are easily taken, set to another, to the few books of copi- the question may arise as to comparability of d illustrations (Price: Dame Fashion; the Kroeber data for 1844-1919, and those of and Boehn), and to the engravings, draw- Richardson for the remaining years. In none of d portraits by fashionable painters of the six measures as tabulated or graphed, is ble women. In these different sources there any large offset between 1843 and 1844, possible to find many plates from the and only one (No. 5) between 1919 and 1920. How- urrier des Dames, previously used by ever, to see if there were any personal equation for 1844-68, and other elegant litho- of measurement, 1844-46 and 1919 were measured issued in monthly or quarterly series on new data by Richardson, and a comparison with "Wiener Moden," the "Galerie des Modes," the Kroeber measures is herewith appended (table 1). TABLE 1 Comparability of Observers* 1844 1845 1846 1919 Dimen. K R K R R K R 2 . ... 97.910 98.223 97.510 97.311 98.29 98.010 84.210 89.418 3 .... 28.91 28.817 27.910 26.910 28.49 28.89 24.110 25.018 4 1 14.61? [ ..... ... 14.323 l 14.19 13.110 l 13.l8 12.010 | 14.23 14.9 18 5 1 57. 58.2 5946 1458.610 57134 62.87 33.21 12.518 7 8 . 1 ... . 9.713 8.44 9.810 8.37 8.89 13.29 13.918 8 20.310 .... 19.220 19.78 l 9.29 18.76 19.38 1 2.98 10.218 *Superior figures to the right are the number of pictures measured from which the per- centage average s here given are derived . * 0 (Y) * ~~~~~~~~~~~~~~~~~~~00 0 '\1 0 0 * 0~~~~~~ 0 1 0 0~~~~ 0 N 0 0~~~~~~~~ 00 * o- 0 0~~~~~~~~ 0 0 * 0~~~ * 0 0 0~~~~ * 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 * 0 0 o0 0 o 0 * 0 co- 0~~~~~~~~~~~~ 0~~~~~~~~~ (0- ~~~~~~~~~~~~~~~~~~~~~0 0) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 gO 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0 0 0) 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ S~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * ~~~~~~~~~~~~~~~~~~~~~~~~~0 * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 * 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ * 0~~~~~~~~~~~~~~~~~~~~~~~~~~~ * 0~~~~~~~~~~~~~~~~~~~~~~~~~~ *0* ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ND- 0~~~~~~~~~~ It 0~~~~~~~~ (0 0~~~~~~~~~~~~~~~~~~~~~~ 0 * 0~~~~~~~S * It) 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 * 0 0~~~~~~~~~~~~~~~~~~~~ * 0 -~~~~~~~~~~~~~~~~~~~ 0~~~~~~~~~~~~~~~~~~ 02 0 bD~*0 0~~~~~~~~~~~~~~~~~~~~~~~ .0 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 * 0 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0 * 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~r * 0 ~~~~~~~~~~~~~~~~~~~~~ 0~~~~~~~~~~~ 0 0~~~~~~~~~~~~~~~~~~~~~~~ * 0 -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\ _ 116 ANTHROPOLOGICAL RECORDS The two sets of data for 1844-46 seem similar less fashion variability, hence the two enough. In only one case, dimension 5 for 1846, measurements coincide more closely. is the difference of the two sets of averaged On the whole, it seems that the sets measures more than 1.5. measurements of the two observers are The 1919 comparisons, however, show a start- ciently alike to be treated as parts of ling difference of 20.7 in dimension 5: 33.2 series. against 12.5. Here Kroeber's data were taken Another problem concerns the differen solely from the March number of Vogue.3 Four tween the signed French and the anonymous of these dresses had trains, six did not. His can designs, as shown in any one magazine percentages run (* denoting train): 33; 22; 134 also what difference might exist between or *33; 10; 19 or *67; 16 or *60; 19 or *64; magazines themselves. Table 2 shows suc 17; 17; 9. He used the higher train widths for tions. his average of 33.2. The Richardson series for In only one case does a range as great 1919 was taken from Vogue of December 1918, and occur: No. 4, 1919, 17.5 and 13.2. This TABLE 2 Comparability of Sources 1919 1919 1919 1920 1920 19 Vogue Costume Costume Vogue Costume CoB Dimen. ord. Royal Royal ord. Royal Ra models ord. mod. French models models ord. mod. French (Dec., 1918) (Apr., June) (Apr., June) (Feb., May) (Apr., June) (Apr. 2 .... 89.7 91.0 87.7 83.3 83.0 79. 3 .... 24.9 26.0 25.1 26.4 26.1 27. 4 .... 17.5 13.2 14.0 16.4 15.3 13 5 .... 12.6 10.0 14.9 15.2 17.6 15. 7 .... 13.8 13.0 14.8 14.5 14.9 14 8 .... 11.1 9.0 10.5 12.2 12.5 13. *Includes one French model from Vogue. from Costume Royal for April and June, 1919. one and a half times the standard deviati All eighteen of these illustrations have no the period 1916-22: 3.6, 0.8, 2.1, 2.0, 2 trains, and they average 12.5. The Kroeber 4.6, mean 2.8. There is a suggestion that average for the ten widths minus trains is French models of 1919 are closer to the o 17.5. This is much nearer the Richardson value ones of 1920: compare dimensions 2, 7. T of 12.5, and it fits well between Kroeber 1918, reflect the fact that the French models u 20.3, and Richardson 1920, 16.7. 1919 and 1920 are late spring modes, and The other larger discrepancies for 1919 are ready pointing the way to fashions of the in skirt length: Kroeber 84.2, Richardson 89.4; ing winter. and in decolletage width: Kroeber 12.9, Richard- Our conclusion is that any differenceb son 10.2. These we cannot explain, unless it the Kroeber and Richardson measurements, be that since 1919 falls in a period of high in- tween models of different magazines, or d dividual variability, as shown by the standard of French and American design, seems to be deviations, any two samples of ten and eighteen on the whole, than the fashion changes fro plates, respectively, might well differ as much. to year. This also holds as regards the The period adjacent to 1844-46 is one of much in illustrations from drawings to photogra 3 living models in the second decade of the Kroeber, 244. century, on which Kroeber has previously g 4The smaller number given for the trained some sample data. 5 dress is the width of the skirt exclusive of the train. These measures are from the manuscript 5P. 244. data. II. MEASUREMENT DATA ta obtained from the several thousand which is therefore the value assigned to dimen- ta made by the two authors, respec- sion No. 2 for 1806. Two of the eleven plates ty years ago and more recently, will did not show the full width of skirt; the nine en in the form of means for each di- that did, yielded 25.2, 46.3, 13.0, 57.0, 35.6, proportion in each year studied. 50.4, 48.5, 35.1, 35.2 per cent, whose mean, 38.4, tance, for 1806, eleven fashion plates is the dimension No. 5 value for 1806. The indi- ations were found which showed length vidual raw and percentaged measurements, as here wthout impairment. These, in terms of cited in illustration, are being preserved, and length of figure as defined, ran to the latter have of course been used in calcula- 5, 98.5, 97.3, 98.0, 97.1, 97.4, 98.5, tion of the standard deviations or variability 5, 98.4 per cent. (In other words, the coefficients presented and discussed in section te measured showed a "length of figure" VII; but they are not printed on account of th down] of 119 mm., a length to bottom cost. Their means for each dimension each year of 113 mm.; the second is 95.0 per cent replace them. rst. The second plate happened also to These means, on which all our other quantita- figure length 119 mm.; but the skirt tive expressions rest, are given in tables 3 6, giving 97.5 per cent. And so on.) and 4. The first of these two tables covers of these eleven percentages is 97.7, the period from 1787 to 1936, in which almost TABLE 3 Ratio of Dress Diameters to Height of Figure: Year-by-Year Means, 1787-1936 Year ~~2 3 4 5 7 8 Year___ |L.Sk. L.Wai. L.Dec. W.Sk. W.Wai. W.Dec. 1787 .... 98.73 27.43 7.23 56.33 11.21 7.91 1788 .... 96.14 20.84 8.92 51.12 16.51 1789 .... 96.97 23.26 10.37 55.15 14.94 10.73 1790 .... 98.68 25.88 7.97 51.07 9*95 7.61 1791 .... 98.34 25.44 11.94 53.23 9.04 8.53 1792 .... 98.51 25.42 9.81 58.42 9.32 1793 .... 98.13 23.74 5.33 52.13 10.64 1794 .... 100.02 18.82 9.92 43.32 14.42 9.72 1795 .... 96.98 20.1' 12.97 52.66 12.36 7.02 1796 .... 98.25 20.64 10.24 43.83 18.0' 1797 .... 99.43 20.72 13.32 49.33 14.92 8.01 1798 .... 98.68 19.99 11.58 44.410 10.59 10.38 1799 .... 97.712 19.413 11.612 42.77 11.210 9.510 1800 .... 98.111 18.111 10.511 42.911 12.08 11.96 1801 .... 96.47 17.96 15.54 42.26 16.02 11.92 1802 .... 100.02 19.442 16.31 59.61 11.32 14.91 1803 .... 97.712 19.012 12.610 38.58 12.88 13.08 1804 .... 95.110 19.410 12.38 44.79 13.55 14.34 1805 .... 97.312 18.311 14.410 35.07 14.85 14.36 1806 .... 97.711 17.911 12.59 38.59 14.36 12.8E 1807 .... 96.518 18.318 11.015 37.815 15.28 14.76 1808 .... 95.610 18.97 11.89 32.67 14.13 13.86 1809 95.815 17.41' 11.714 24.813 15.113 13.713 1810 .... 95.67 21.06 13.57 24.27 12.95 12.34 1811 94.48 21.77 13.36 27.28 13.57 15.05 1812 .... 92.74 20.94 13.03 29.44 13.42 17.02 1813 88.03 19.23 11.13 28.93 14.43 19.83 1814 ... 94.2' 19.8' 13.23 31.3' 12.8' 17.23 1815 ... 92.95 17.35 11. 94 45.85 13.63 15.25 1816 ... 91.63 17.23 12.03 39.83 12.02 16.62 1817 ... 94.72 20.12 16.52 42.92 13.5' 9.62 1818 ... 93.93 20.63 11.32 38.63 11.32 16.42 1819 ... 90.1' 15.51 13.21 41.1' 10.1' 16.3' [117] 118 TABLE 3 (Continued) Year ~2 3 4 5 7 8 Year L.Sk. L.Wai. L.Dec. W.Sk. W.Wai. W.Dec. 1820 .... 92.23 21.43 13.22 3?.03 10.03 17.22 1821 .... 96.43 21.73 10.12 42.73 11.22 17.83 1822 .... 1823 .... 96.93 24.43 12.12 46.33 10.38 18.32 1824 .... 92.52 21.52 13.82 38.12 12.42 18.12 1825 .... 95.51 25.41 10.91 48.61 10.11 22.51 1826 .... 95.02 25.22 10.82 48.52 10.42 17.02 1827 .... 92.86 28.45 12.25 53.26 11.26 23.23 1828 .... 90.410 26.710 12.28 49.610 10.06 21.39 1829 .... 93.93 26.73 12.21 55.63 11.33 21.42 1830 .... 93.79 27.99 12.24 48.78 12.18 24.72 1831 .... 94.84 23.74 12.13 44.24 13.53 21.21 1832 .... 90.8 26.4 12.9 54.4 11.0 23.8 1833 1834 .... 89.44 27.04 14.03 64.33 9.84 23.93 1835 .... 92.99 28.09 12.77 67.09 9.49 17.07 1836 .... 95.914 26.913 12.914 70.012 9.49 17.9'o 1837 .... 95.711 27.310 13.49 62.38 9. 18 19.16 1838 .... 97.010 28.910 14.08 70.98 9.15 19.96 1839 .... 97.114 28.713 14.813 63.613 9.510 20.710 1840 .... 97.022 28.421 14.320 63.91'5 9.02O 20.618 1841 .... 96. 92 29.1P3 14.827 64.617 8.817 19.31 9 1842 .... 96.917 29.213 14.516 60.014 9.19 18.08 1843 .... 97.324 29.021 14.422 60.116 9.116 20.719 1844 .... 98.210 28.810 14.410 58.27 9.lO 19.210 1845 .... 97.510 27.910 14.19 59.46 8.44 19. 88 1846 .... 98.39 28.49 13.18 57.34 8.37 18.76 1847 .... 98.410 28.99 14.89 64.84 8.86 19.68 1848 .... 98.010 27.810 13.410 59.65 8.58 20.010 1849 .... 97.910 28.710 13.39 62.76 8.48 20.09 1850 .... 97.810 28.610 12.710 64.26 8.38 20.710 1851 .... 98.710 29.410 13.910 61.37 8.49 21.28 1852 .... 97.610 27.010 14.19 70.Q3' 8.310 21.49 1853 .... 98.110 27.710 12.810 70.26 7.88 21.29 1854 .... 97.910 27.010 14.110 79.37 8.49 20.68 1855 .... 98.210 27.910 13.310 83. 04 9.08 21.010 1856 .... 98.39 27.79 13.47 89.24 8.65 19.19 1857 .... 98.410 26.710 13.910 86.25 8.0 5 19.610 1858 .... 99.610 26.810 15.210 100.38 7.97 18.810 1859 .... 100.010 25.39 14.410 115.69 7.83 18.38 1860 .... 99.810 24.810 12.310 107.17 7.65 18.19 1861 .... 100.010 24.910 12.310 104.39 8.04 17.810 1862 .... 99.610 24.110 13.210 96.19 7.68 17.99 1863 .... 98.710 24.97 13.19 101.69 9.08 17.19 1864 .... 99.510 23.97 13.59 100.11 8.55 18.19 1865 .... 99.810 22.89 12.79 108.61 8.67 17.58 1866 .... 99.89 22.49 12.810 99.82 8.29 18409 1867 .... 97.910 21.210 11.79 98.79 8.19 16.79 1868 .... 98.810 22.09 12.77 88.410 9.08 16.18 1869 .... 100.010 21.810 13.69 85.58 9.04 16.07 1870 .... 99.19 22.29 12.08 88.06 9.24 18.56 1871 .... 99.310 22.09 13.06 74.98 9.03 16.45 1872 .... 99.lo 22.910 15.06 77.68 11.03 16.72 1873 .... 99.210 24.510 13.85 84.810 10.16 18.33 1874 .... 99.29 22.210 14.17 84.56 9I53 15. 04 1875 . . .. 100.010 22.39 14.17 79.07 10.45 17.0w 1876 ..... 99.210 23.610 13.47 84.88 9.54 13.52 1877 ..... 98.710 23.810 13.56 76.410 8.76 13.76 1878 ..... 99.010o 24.810 14.56 7Q.98 8.96 14.68 1879 . . .. 98.710 26.110 13.36 62.0o5 8.88 15.08 RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 119 TABLE 3 (Concluded) Year 2 3 4 5 7 8 Year___| L.Sk. L.Wai. L.Dec. W.Sk. W.Wai. W.Dec. 1880 .... 98.810 27.610 15.46 68.86 8.88 14.18 1881 .... 97.710 27.610 14.27 52.37 8.59 14.76 1882 .... 96.610 26.010 12.88 56.08 7.87 15.37 1883 .... 96.910 26.010 12.89 54.77 8.68 17.37 1884 .... 96.410 26.210 13.19 52.29 8.26 14.46 1885 .... 97.010 27.410 14.0 7 56.06 8.78 15.25 1886 .... 95.89 27.310 14.99 56.69 8.910 14.88 1887 .... 95.610 27.210 12.98 50.99 8.38 14.28 1888 .... 95.710 27.610 14.19 57.88 8.310 13.1i 1889 .... 96.710 27.710 13.710 51.59 9.68 13.29 1890 .... 97.310 28.210 14.18 50.210 8.510 13.58 1891 .... 97.310 28.310 14.49 53.79 9.210 12.6r 1892 .... 97.410 28.810 13.47 51.110 9.210 14.39 1893 .... 98.810 27.010 13.68 55.010 9.39 13.38 1894 .... 98.210 28.810 14.09 55.510 9.210 14.09 1895 .... 98.710 27.410 14.09 60.79 8.67 15.07 1896 .... 99.210 27.910 14.310 68.38 9.67 15.29 1897 .... 99.910 28.910 14.410 60.010 8.69 15.810 1898 .... 99.810 29.510 14.710 53.010 8. 18 11.99 1899 .... 100-010 29.710 14.69 65.310 9.39 12.510 1900 .... 99.310 30.510 15.110 52.510 8.710 13.49 1901 .... 99.710 30.510 12.510 64.810 9.410 13.310 1902 .... 100.010 30.110 13.19 58.910 9g98 11.16 1903 .... 100.010 32.69 15.29 50.410 9.68 13.07 1904 .... 100.010 32.310 14.09 56.510 9.98 14.87 1905 .... 100.010 30.310 14.610 53.710 9.29 15.38 1906 .... 99.610 28.810 16.25 56.010 9.58 11.25 1907 .... 99.610 28.010 13.39 51.210 9.76 12.37 1908 ' 99.310 25.49 11.79 49.010 10.98 12.98 1909 .... 99.710 24.310 14.67 38.410 12.89 12.110 1910 .... 99.210 25.210 13.38 32.99 11.79 13.09 1911 .... 98.710 26.110 14.26 23.210 12.09 12.26 1912 ' 98.310 24.310 13.49 27.410 13.29 11.57 1913 ..... 92.69 25.510 15.47 33.77 13.69 13.310 1914 .... 91.89 25.310 14.47 29.17 13.910 15.24 1915 .... 91.110 24.49 16.26 46.110 13.79 11.25 1916 .... 84.310 25.09 16.49 49.110 12.69 12.36 1917 .... 88.110 24.38 14.89 55.74 13.09 11.47 1918 .... 85.310 24.210 13.79 20.38 13.410 10.68 1919 .... 84.210 24.110 14.29 33.210 13.29 12.98 1920 .... 82.217 26.517 15.017 16.817 14.717 12.917 1921 .... 83.121 27. 921 13.421 26.021 15.221 14.721 1922 .... 86.014 33.314 14.314 26.414 16.514 13. 114 1923 .... 90.710 30.910 13.310 22.410 15.410 14.810 1924 .... 85.613 38.013 13.013 21.313 15.913 13.313 1925 .... 77.010 35.510 12.610 22.010 15.610 12.610 1926 .... 72.211 31.911 15.811 16.511 14.211 12.3" 1927 .... 69.711 37.911 13.011 18.011 13.811 11.2" 1928 .... 70.010 36.210 15.010 21.210 14.710 12.110 1929 .... 71.220 32.520 13.020 26.020 13.420 11.110 1930 .... 89.020 28.120 14.620 25.920 16.520 11.720 1931 .... 95.11'3 26.812 14.912 26.012 13.012 12.812 1932 .... 97.310 24.410 11.410 25.810 11.010 10.310 1933 .... 98.310 25.010 11.210 25.910 13.110 13.210 1934 ..... 99.310 25.010 9.310 26.210 13.7'0 11.810 1935 ..... 98.2'7 25.216 11.119 38.716 9912 11.412 1936 ..... 98.82 5 25.32 3 11 .822 33.821 10.7'7 9.816 120 ANTHROPOLOGICAL RECORDS every consecutive year is represented and for fuller period 1787-1926. Table 7 combine most years the data are reasonably sufficient. by pairs into ten-year means. It is, ho (See figs. 1-2, pp. 114-115.) The second table not a mere replica of the last part of ta the preceding one hundred and eighty-two years because the absolute dates are different: from 1605 to 1786, where the data are both dis- table 5 treats the years 1881-90 as a uni continuous and fewer. In both these tables, the table 7 lumps 1877-86. small superior figures at the right indicate the Next comes the "trend" or moving aver, number of cases measured. table 8. Here the value for any given d' Next follow combinations of these year-by- for 1840 is the average of the values for year means into the averages for longer blocks 1839, 1840, 1841, 1842 as found in table of years. Table 5 covers the whole three hun- 1841, of 1839, 1840, 1841, 1842, 1843. dred and thirty-two years in ten-year periods. way, the "exceptionalness" of any single (See figs. 3, 4, 5.) This seemed a better in- minimized by its being merged with the ye terval than five-year averagings on account of either side, and a smoother curve results the irregularity of the data before 1787. It instance, from 1846 to 1855 skirt length should be noted that these values, and all other all intents and purposes at a stand-still, averages of the year-by-year means, are un- individual year 1852 comes out, in the a weighted for number of cases. For instance, for ments that happened to be made, somewhat the decade 1641-50, only four dated illustra- than any other in that decade: 97.6 as tions were found: three for 1641 and one for nine others ranging from 97.8 to 98.7. I 1647. In dimension 5, skirt width, the three doubtful whether a fluctuation so small for 1641 averaged 72.2, the one for 1647 was is either significant or reliable. The m 78.0. The mean of these two values is 75.1, average, which stays at 98.0, 98.1, 98.2 which is the one that will be found in table 5 decade, probably gives a truer picture of. for this decade; whereas the weighted average events. On. account of its much greater a would be 73.7. ness, its variations are also much more Table 6 gives the five-year means for the grasped, when they do become appreciable. TABLE 4 Year-by-Year Means, 1605-1786 Year 2 3 4 5 7 8 1605 .... 100.0 23.5 7.0 76.4 1610 .... 97.12 19.42 3.62 68.1 11.62 5.82 1613 .... 95.9 22.5 8.5 78.9 14.1 8.5 1617 . .. 100.0 21.7 2.8 13.0 5.7 1622-27.. 100.02 25.22 3.72 13.6 7.9 1625 .... 26.5 16.3 16.3 1628 .... 96.4 28.6 2.4 53.5 14.3 4.8 1629 .... 97.44 25.92 7.64 41.7 3 11.3 6.73 1630 .... 100.0 25.9 2.8 11.1 5.5 1631 .... 100.02 23.72 4.82 54.52 6.32 1633 .... 99.2 3 24.3 3 13.23 49.62 14.8 16.33 1634 .... 97.83 22.63 6.73 50.43 7.63 1635 .... 100.04 24.14 9.74 52.6 13.04 1636 .... 100.02 21.22 13.1 35.6 16.2 1638 .... 100.0 24.1 11.1 63.0 14.8 1639 .... 100.03 23.53 14.13 52.23 15.53 1640 .... 99.O3 23.94 14.2 46.44 14.22 1641 .... 100.03 26.12 10.13 72.23 15.8 13.73 1647 .... 100.0 27.1 3.1 78.0 9.3 1656-57.. 100.0 36.6 10.0 87.5 17.5 1658 .... 100.0 27.4 6.8 82.2 11.7 10.9 1659 .... 100.0 29.2 9.0 126.2 16.2 22.2 1660 ..... 100. 02 29.72 12.22 88.92 18.82 1666 ..... 100.0 32.2 11.3 58.0 1667 ..... 98.4 6.9 45.0 1668 ..... 100.0 30.2 15.1 75.5 22.6 RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 121 TABLE 4 (Continued) Year 2 3 4 5 7 1678 .... 97.5 33.3 15.4 33.4 10.2 1679 .... 97.62 23.3 11.6 40.02 1680 .... 96.2 31.5 14.8 40.8 9.3 16.7 1683 97.32 29.12 12.3 30.12 12.42 1685 .... 100.0 13.1 63.7 11.9 1686 .... 99.2 29.1 48.7 1710-16.. 99.49 24.97 13.75 52.66s 13.64 13.63 1711 .... 96.5 29.0 15.3 73.8 11.9 14.5 1714 .... 97.52 25.32 10.0 25.5 10.0 10.0 1715 .... 100.02 26.72 46.42 17.4 13.0 1716-18.. 98.72 25.72 12.1 58.32 10.7 1717 .... 96.4 21.4 14.3 50.0 1719 .... 100.0 46.5 1720 .... 100.02 29.4 15.02 97.0 14.7 14.7 1721 .... 89.6 62.3 1726 .... 100.0 26.6 18.3 15.0 1728 .... 100.0 10.5 57.9 1731 .... 92.82 12.9 61.32 1734 .... 98.03 26.13 15.3a 76.4 12.52 13.2 1735 .... 96.22 28.83 13.52 66.6 12.8 13.8 1738 .... 96.45 27.95 16.45 54.24 16.7 10.93 1739 .... 97.0 28.2 13.4 13.4 23.9 1740 .... 99.0 28.6 14.3 64.3 13.4 11.6 1742 . 93.82 25.62 7.7 134.5 10.8 1743 27.5 13.3 14.2 10.0 1744 .... 100.0 24.3 14.2 70.0 11.4 1744-45.. 96.4 12.5 91.0 7.1 1749 .... 95.6 28.6 14.2 102.9 12.9 20.0 1750 .... 97.22 25.2 15.42 101.0 18.7 1751 .... 96.5 30.1 15.1 100.7 12.3 21.2 1752 .... 98.52 26.22 14.42 44.9 14.3 12.3 1753 .... 93:52 26.92 10.9 31.82 10.0 1755 .... 95.22 30.33 16.7 59.03 13.0 8.8 1757 .... 91.4 27.5 14.7 58.4 14.3 1758 .... 97.2 31.2 12.2 39.7 1759 .... 98.0 30.2 17.4 80.3 14.5 13.0 1761 .... 89.5 25.0 16.8 65.1 14.5 12.2 1762-64.. 92.3 25.0 15.62 72.3 12.72 11.5 1765 .... 94.53 27.33 13.13 65.9 12.32 10.9.3 1766 .... 97.6 14.9 35.6 1767 .... 95.5 11.5 48.4 1769 .... 95.5 31.42 15.22 59.0 9.6 1770 .... 98.4 22.1 11.6 34.8 8.8 1771 .... 98.5 30.8 19.2 40.0 1772 .... 95.0 28.9 16.5 101.0 16.5 1773 .... 91.87 24.86 11.86 46.47 13.6 9.82 1775 .... 96.55 26.32 13.85 57.64 12.9 6.8 1776 .... 95.93 27.93 16.03 72.52 10.32 12.42 1777 .... 95.45 26.84 12.85 77.03 10.73 9.94 1778 .... 92.42 33.52 18.42 55.32 10.9 8.7 1779 .... 92.32 30.82 16.52 41.62 1780 .... 9O.33 26.63 15.53 57.93 8.6 8.32 1781 . . .. 92.22 23.9 16.4 57.5 10.5 1784 . . .. 93.3' 24.4' 11.03 50.02 9.2 11.02 1786 ... 97.4 23.7 8.8 36.8 122 ANTHROPOLOGICAL RECORDS for example the moving average of waist length brings out 1926 as the peak of the curve for 1923-32, which runs, slightly rounded from the actual figures this is, however, low table 8: 33, 34, 35, 36, 35, 34, 33, 30, 28, 26, for the series of,six from 1924 to 1929, Il9Op 2P op 6p ri0 o7a0 ~o so 66,o B ePo ISP 21 op Sp 19o00 21o1 1G,OOap r0 Sp0 I 17ZO ~O 10 rOP ~0 I9IWO20 r S op.' 80- - 5 82- - 6 91- - 7- 88- - 8- SB- - ~~~~~~~~~~~9- 4 90-- 10- 01- ~2 - 12- 100- - 18 16- 18- 100- 90- 20- 22- 3o \5 80- 28- 10- so-~~~~~~~~~~~~~~~~~~~~s so-~~~~~~~~~~~~~~~o 82-~~~~~~~~~~~~~~~0 20- Fig. 3. Skirt length and waist length, dimen- Fig. 4. Decolletage depth and skirt sions 2 and 3, by ten-year means, 1605-1936. dimensions 4 and 5, by ten-year means, as against the year-by-year values -of 30, 38, peak is left double and indeterminate b 36, 32, 38, 36, 33, 28, 27, 24 of table 3. The 1924 and 1927. We assume that there smoothed or idealized moving-average series tical techniques expressing the probabil TABLE 5 Ten-Year Averages, 1605-1936* Period 2 3 4 5 78 1605-10 .... 98.63 21.53 5.33 72.3 2 11.6 2 5.8 2 1611-20 ... 98.0O1 22.12 5.72 78.91 13.6 2 7.12 1621-30 ... 98.58 26.47 6.69 47.6' 4 l3.3E, 6.26 1631-40 .... 99.521 23.4 22 10.918 50.517 14.81 13.019 1641-50 .... 100.0' 2. ~ 66 75. 1' 15.81 11.54 1651-60 .... 10O.O5 30.7-9 9.55 96.25 14.02 17.45 1661-70 .... 99.53 31.22 11.13 59.5 3 22.61 1671-80 .... 97.14 29.43 13.9-3 38.1' 9.82 16.71 1681-90 .... 98.8' 29.23 12.72 47.5' 12.2 3 1691-1700 RICHARDSON AND KROEBER: THREE CENURIES OF WOMEN'S DRESS FASHIONS 123 TABLE 5 (Continued) Period 2 3 4 5 7 8 1731-40 .... 96.614 27.913 14.312 60.79 13.88 14.77 1741-50 .... 96.67 26.26 12.97 99.95 12.34 14.04 1751-60 .... 95.810 28.911 14.58 60.210 12.8w 13.95 1761-70 .... 94.89 26.38 14.111 54.47 13.25 10.67 1771-80 .... 94.229 28.524 15.628 50.928 11.910 9.312 1781-90 .... 96.229 24.225 10.124 51.121 12.312 9.58 1791-1800 .. 98. 457 21.258 10.754 48.349 12.247 9.332 1801-10 .... 96.8104 18.898 13.28 7 37.882 14.07 13.6w4 1811-20 .... 92.538 19.435 12.929 36.238 12.528 16.027 1821-30 .... 94.139 25.338 11.827 47.938 11.3 20.526 1831-40 . ... 94.591 27.386 13.580 62.375 10.070 20.563 1841-50 .... 97.7136 28.6125 14.0130 61. 187 8.792 19.6107 1851-60 .... 98.799 27.098 13.796 86.361 8.270 19.990 1861-70 .... 99.398 23.090 12.885 97.186 8.5682 17.488 1871-80 .... 99.199 24.098 14.062 76.475 9.552 15. 452 - 1881-90 .... 96.698 27.1100 13.784 53.883 8.583 14. 672 1891-1900 .. 98.9100 28.7100 14.391 57,598 9.089 13.885 1901-10 .... 99.7100 28.898 13.985 51.299 10.384 12.974 1911-20 .... 89.7103 25.0103 14.888 33.5104 13.3100 12.477 1921-30 ..... 79.5140 33.2104 13.8140 22.6140 15.1140 12.7140 1931-36 ....o 97.885 25.381 11.683 29.479 11.971 11.670 TABLE 6 Five-Year Averages, 1788-1936 Period 2 3 4 5 7 8 1788-91 .... 97.7 *24.5 t9.2 53.3 12.3 8.7 1792-96 .... *98.3 21.7 9.6 50.0 12.9 8.4 1797-1801 .. 98.0 19.2 12.5 44.3 12.9 10.3 1802-06 . ... 97.6 t18.8 *13.6 43.3 13.3 13.9 1807-11 .... 95.6 19.5 12.3 t29.3 *14.1 13.9 1812-16 . ... t91.9 18.9 12.2 35.0 13.2 17.2 1817-21 .... 93.5 '19.9 12.9 40.5 11.2 15.5 1822-26 .... 95.0 24.1 tll.9 45.4 10.8 19.0 1827-31 . ... 93.1 26.7 12.2 50.3 11.6 *22.4 1832-36 .... 92.3 27.1 13.1 63.9 9.9 20.7 1837-41 .... 96.7 28.5 14.3 65.1 9.1 19.9 1842-46 .... 97.6 *28.7 14.1 59.0 8.8 19.3 1847-51 .... 98.2 *28.7 13.6 62.5 8.5 20.3 1852-56 .... 98.0 27.5 13.5 78.4 8.4 20.7 1857-61 .... *99.6 25.7 13.6 *102.7 t7.9 18.5 1862-66 .... 99.5 23.6 13.4 101.2 8.4 17.7 1867-71 .... 99.0 t21.8 12.6 87.1 8.9 16.7 1872-76 .... 99.4 23.1 14.1 82.1 10.1 16.1 1877-81 .... 98.6 26.0 14.2 66.1 8.7 14.4 1882-86 .... t96.5 26.6 13.5 55.1 8.4 15.4 1887-91 .... t96.5 27.8 13.8 52.8 8.8 13.3 1892-96 .... 98.5 28.0 13.9 58.1 9.2 14.4 1897-1901 . . 99.7 29.9 14.3 59.1 8.8 13.4 1902-06 .... *99.9 *30.8 *14.6 55.1 9.6 13.1 1907-11 ....... 99.3 25.8 13.4 38.9 11.4 12.5 1912-16 . . .. 91.6 t-24.9 15.2 37.1 13.4 12.7 1917-21 ....... 84.6 25.4 14.2 30.4 13.9 12.5 1922-26 . . .. 82.3 *33.9 13.8 t-21.7 *15.5 13.2 1927-31 . . .. 1-79.0 32.3 14.1 23.4 14.3 11.8 1932-36 . . .. 98.4 25.0 1-11.0 30.1 11.8 tll .3 *High poi nts 1-tLcow point s. 124 ANTHROPOLOGICAL RECORDS improbability of 1926 really being a year of IqO 20 30 610 810 17g0O0 10 GPO 9a 10AI0 30 610 temporary recession near the climax of the curve, 1 or on the contrary of its low value being merely A 7 a by-product of the unrepresentativeness of the is- 7 small number of pictures measured. But these 15- techniques are cumbersome, and the issues in- 13- volved are small. Whether the waist line over 12- a period of one hundred and fifty years reached /1\ its lowest position (highest percentage) pre- cisely in 1926 or perhaps rather in either 1924 10- or 1927, can be of no great moment. The essen- 9- tial truth seems sufficiently expressed, and 9_ certainly much more vividly, by the moving average. 23- For this reason we have graphed the moving 21- 8 averages in figures 1 and 2 (pp. 114-115); 19- These should be considered our basic diagrams. 17- We have shown in these diagrams also the actual values for each year, from table 3; but these values have been indicated by disparate points 18 or dots, whereas the moving-average values are 1- connected by a continuous line. Earlier than 1787, the data are too scattered 7 for a satisfactory moving average, and it has 5- not been attempted. Its place is partially I I I , I I i l l taken by the means of means in ten-year blocks Fig. 5. Waist width and decolletage, in the first part of table 5. dimensions 7 and 8, by ten-year means, TABLE 7 Ten-Year Averages, 1788-1936, By Decades Ending in 6 Period 2 3 4 5 7 8 1788-96 .... 98.3 23.1 9.4 51.7 12.6 8.5 1797-1806 .. 97.8 19.0 13.1 43.8 13.1 12.1 1807-16 .... 93.7 19.2 12.3 32.2 13.7 15.5 1817-26 .... 94.1 21.8 12.4 42.6 11.0 17.0 1827-36 .... 92.7 26.9 12.6 56.3 10.9 21.6 1837-46 .... 97.2 28.6 14.2 62.0 8.9 19.6 1847-56 .... 98.1 28.1 13.6 70.5 8.5 20.5 1857-66 .... 99.5 24.7 13.3 102.0 8.1 18.1 1867-76 .... 99.2 22.5 13.4 84.6 9.5 16.4 1877-86 .... 97.6 26.3 13.9 60.6 8.6 14.9 1887-96 .... 97.5 27.9 13.9 55.5 9.0 13.8 1897-1906 .. 99.8 30.3 14.4 57.1 9.2 13.2 1907-16 ... 95.5 1 25.4 1 14.3 - 38.0 - 12.4 1 12.6 1917-26 ... . 83.4 29.7 14.0 26.1 14.7 12.9 1927-36 ...... 88.7 28.6 12.5 26.8 13.0 11.5 - RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 125 TABLE 8 Five-Year Moving Average of Dress Diameters. 1788-1934 Year ~ 2 3 4 5 78 Year | L.Sk. LWai.| L.Dec. W.Sk. W.Wai. W.Dec. 1788 .... 97.5 24.2 t8.6 50.1 12.5 1789 .... 97.7 24.5 9.2 53.3 12.3 8.7 1790 .... 97.7 24.1 9.8 53.8 11.9 8.9 1791 .... 98.1 *24.7 9.0 *54.0 10.7 8.9 1792 .... *98.7 23.8 9.0 51.6 10.6 8.6 1793 .... 98.4 22.7 10.0 51.9 11.1 8.4 1794 .... 98.3 21.7 9.6 50.0 12.9 8.4 1795 .... 98.5 20.8 10.3 48.2 14.0 t8.2 1796 .... 98.6 20.0 11.6 46.7 14.0 8.8 1797 .... 98.2 20.1 11.9 46.6 13.4 8.7 1798 .... 98.4 19.7 11.4 44.6 13.3 9.9 1799 .... 98.0 19.2 12.5 44.3 12.9 10.3 1800 .... 98.2 18.9 13.1 46.4 12.2 11.7 1801 .... 98.0 18.8 13.3 45.2 12.7 12.2 1802 .... 97.5 18.8 13.4 45.6 13.1 13.2 1803 .... 97.3 18.8 *14.2 44.9 13.7 13.7 1804 .... 97.6 18.8 13.6 43.3 13.3 13.8 1805 .... 96.9 18.6 12.6 38.9 14.1 11.8 1806 .... 96.4 18.6 12.4 37.7 14.4 14.0 1807 . .. 96.6 18.2 12.3 33.7 *14.7 13.9 1808 .... 96.2 18.7 12.1 31.6 14.3 13.5 1809 .... 95.6 19.5 12.3 29.3 14.1 13.9 1810 .... 95.8 20.0 12.7 27.6 13.8 14.4 1811 .... 93.3 20.0 12.5 t26.9 13.8 15.6 1812 .... 93.0 20.5 12.8 28.2 13.4 16.3 1813 .... 92.4 19.8 12.5 32.5 13.5 16.8 1814 .... t91.9 18.9 12.2 35.0 13.2 17.2 1815 .... 92.3 18.7 12.9 37.7 13.3 15.7 1816 .... 93.5 19.0 13.0 39.7 12.6 15.0 1817 .... 92.6 t18.1 13.0 41.6 12.1 14.8 1818 .... 92.5 19.0 13.2 39.9 11.4 15.2 1819 .... 93.5 19.9 12.9 40.5 11.2 15.5 1820 .... 93.2 19.8 12.0 39.9 10.7 16.9 1821 .... 93.9 20.8 12.2 41.8 10.4 17.4 1822 .... 94.5 22.3 12.3 41.0 11.0 17.9 1823 .... 95.3 23.3 tl1.7 43.9 11.0 19.2 1824 .... 95.0 24.1 11.9 45.4 10.8 19.0 1825 .... 94.5 25.0 12.0 46.9 10.9 19.8 1826 .... 93.2 25.4 12.0 47.6 10.8 20.4 1827 .... 93.5 26.5 t11.7 51.1 10.6 21.1 1828 .... 93.2 27.0 11.9 51.1 11.0 21.5 1829 .... 93.1 26.7 12.2 50.3 11.6 22.4 1830 .... 92.7 26.3 12.3 50.5 11.6 22.5 1831 .... 93.3 26.2 12.4 50.7 12.0 22.8 1832 .... 92.2 26.3 12.8 54.9 11.6 *23.4 1833 .... 92.0 26.2 12.9 57.5 10.9 21.5 1834 .... 92.3 27.1 13.1 63.9 9.9 20.7 1835 .... 93.5 27.3 13.3 65.9 9.4 19.5 1836 .... 94.2 27.6 13.4 66.9 9.4 19.6 1837 .... 95.7 28.0 13.6 66.8 9.3 18.9 1838 .... 96.5 28.0 13.9 66.1 9.2 19.6 1839 .... 96.7 28.5 14.3 65.1 9.1 19.9 1840 .... 97.0 *28.9 14.5 64.6 9.1 19.7 1841 ..... 97.0 *28.9 *14.6 62.4 9.1 19.9 1842 ..... 97.3 *28.9 14.5 61.4 9.0 19.6 1843 ..... 97.4 28.8 14.4 60.5 8.9 19.4 1844 ..... 97.6 28.7 14.1 59.0 8.8 19.3 1845 ..... 97.9 28.6 14.2 60.0 8.7 19.6 *High points. 1-Low points. 126 ANTHROPOLOGICAL RECORDS TABLE 8 (Continued) Year 2 3 4 5 7 8 Year | L.Sk. L.Wai. L.Dec. W.Sk. W.Wai. W.Dec. 1846 .... 98.1 28.4 14.0 59.9 8.6 19.5 1847 .... 98.0 28.3 13.7 60.8 8.5 19.6 1848 .... 98.1 28.5 13.5 61.7 8.5 19.8 1849 .... 98.2 28.7 13.6 62.5 8.5 20.3 1850 .... 98.0 28.3 13.5 63.6 8.4 20.7 1851 .... 98.0 28.3 13.4 65.7 8.2 20.9 1852 .... 98.0 27.9 13.5 69.1 8.3 21.0 1853 .... 98.1 27.8 13.6 72.8 8.4 21.1 1854 .... 98.0 27.5 13.5 78.4 8.4 20.7 1855 .... 98.2 27.4 13.5 81.6 8.4 20.3 1856 .... 98.5 27.2 14.0 87.6 8.4 19.8 1857 .... 98.9 26.9 14.0 94.9 8.3 19.4 1858 .... 99.2 26.3 13.8 99.7 8.0 18.8 1859 .... 99.6 25.7 13.6 102.7 7.9 18.5 1860 .... *99.8 25.2 13.5 104.7 t7.8 18.2 1861 .... 99.6 24.8 13.1 *104.9 8.0 17.8 1862 .... 99.5 24.5 12.9 101.8 8.1 17.8 1863 .... 99.5 24.1 13.0 102.1 8.3 17.7 1864 .... 99.5 23.6 13.1 101.2 8.4 17.7 1865 .... 99.1 23.0 12.8 101.8 8.5 17.5 1866 .... 99.2 22.5 12.7 99.1 8.5 17.3 1867 .... 99.4 22.4 12.7 96.2 8.6 16.9 1868 .... 99.1 21.9 t12.6 92.1 8.7 17.1 1869 .... 99.0 t21.8 t12.6 87.1 8.9 16.7 1870 .... 99.3 22.2 13.3 82.9 9.4 16.7 1871 .... 99.4 22.7 13.5 82.2 9.7 17.2 1872 .... 99.2 22.8 13.6 82.0 9.8 17.0 1873 .... 99.4 22.8 14.0 80.2 10.0 16.7 1874 .... 99.4 23.1 14.1 82.1 10.1 16.1 1875 .... 99.3 23.3 13.8 81.9 9.6 15.5 1876 .... 99.2 23.3 13.9 79.1 9.4 14.8 1877 .... 99.1 24.1 13.8 74.6 9.3 14.8 1878 .... 98.9 25.2 14.0 72.6 8.9 14.2 1879 .... 98.6 26.0 14.2 66.1 8.7 14.4 1880 .... 98.2 26.4 14.0 62.0 8.6 14.7 1881 .... 97.7 26.7 13.7 58.8 8.5 15.3 1882 .... 97.3 26.7 13.7 56.8 8.4 15.2 1883 .... 96.9 26.6 13.4 54.2 8.4 15.4 1884 .... 96.5 26.6 13.5 55.1 8.4 15.4 1885 .... 96.3 26.8 13.5 54.1 8.5 15.2 1886 .... t96.1 27.1 13.8 55.7 8.5 14.3 1887 .... 96.2 27.4 13.9 55.6 8.8 14.1 1888 .... 96.2 27.6 13.9 53.4 8.7 13.8 1889 .... 96.5 27.8 13.8 52.8 8.8 13.3 1890 .... 96.9 28.1 13.9 52.9 9.0 13.3 1891 .... 97.5 28.0 13.8 52.3 9.2 13.4 1892 .... 97.8 28.2 13.9 53.1 9.1 13.5 1893 .... 98.1 28.1 13.9 55.2 9.1 13.8 1894 .... 98.5 28.0 13.9 58.1 9.2 14.6 1895 .... 99.0 28.0 14.1 59.9 9.1 14.7 1896 .... 99.2 28.5 14.3 59.5 8.8 14.4 1897 .... 99.5 28.7 14.4 61.5 8.8 14.1 1898 .... 99.6 29.3 14.6 59.8 8.9 13.8 1899 .... 99.7 29.8 14.3 59.1 8.8 13.4 1900 ..... 99.8 30.1 14.0 58.9 9.1 12.4 1901 . . .. 99.8 30.7 14.1 58.4 9.4 12.7 1902 . . .. 99.8 *31.2 14.0 56.6 9.5 13.1 1903 . . .. *99.9 *31.2 13.9 56.9 9.6 13.5 1904 ..... *99.9 30.8 14.6 55.1 9.6 13.1 *High points. tLow points. RICHARDSON AND EROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 127 TABLE 8 (Concluded) 2 3 4 5 7 8 Year L.Sk. L.Wai. L.Dec. W.Sk. W.Wai. W.Dec. 1905 .... 99.8 30.4 14.7 53.5 9.6 13.3 1906 .... 99.7 29.0 14.0 53.3 9.8 13.3 1907 .... 99.6 27.4 14.1 49.7 10.4 12.8 1908 .... 99.5 26.3 13.8 45.5 10.9 12.3 1909 .... 99.3 25.8 13.4 38.9 11.4 12.5 1910 .... 99.0 25.1 13.4 34.2 12.1 12.3 1911 .... 97.7 25.1 14.2 31.1 12.7 12.4 1912 .... 96.1 25.3 14.1 29.3 12.9 13.0 1913 .... 94.5 25.1 14.7 31.9 13.3 12.7 1914 .... 91.6 24.9 15.2 37.1 13.4 12.7 1915 .... 89.6 24.9 *15.4 43.1 13.4 12.7 1916 .... 88.1 24.6 15.1 40.5 13.3 12.1 1917 .... 86.6 t24.4 15.1 41.3 13.2 11.7 1918 .... 84.8 24.8 14.8 35.4 13.4 12.0 1919 .... 84.6 25.4 14.2 30.8 13.9 12.5 1920 .... 84.2 27.2 14.1 24.5 14.6 12.8 1921 .... 85.2 28.5 14.0 25.0 15.0 13.7 1922 .... 85.5 31.3 13.8 22.6 15.5 13.8 1923 .... 84.5 33.1 13.3 23.6 *15.7 13.7 1924 .... 82.3 34.9 13.8 21.7 15.5 13.2 1925 79.0 34.8 13.5 20.0 15.0 12.8 1926 .... 74.9 *35.9 13.9 t19.8 14.8 12.3 1927 .... t72.0 34.8 13.9 20.7 14.3 11.9 1928 .... 74.4 33.3 14.3 21.5 14.5 11.7 1929 .... 79.0 32.3 14.1 23.4 14.3 11.8 1930 .... 84.5 29.6 13.8 25.0 13.7 11.6 1931 .... 90.2 27.4 13.0 25.9 13.4 11.8 1932 .... 95.8 25.9 12.3 26.0 13.5 12.0 1933 97.6 25.3 11.6 28.5 12.1 11.9 .1934 98.4 25.0 tll.0 30.1 11.7 tll.3 *High points. tLow points. III. DESCRIPTIVE HISTORY OF THE PROPORTIONS OF DRESS dimension of dress appears to have a farthingale (vertugarde, crinoline), which is less independent history. At least it very wide about 1570,8 is completely out by 1625- nsidered independently. It seems ad- 30,7 when skirts attain a relative minimum of to treat the histories of the six di- width. Increasing diameters then follow until in two ways: first, descriptively, here- about 1660. This is reflected in the court styles again quantitatively, in the following painted by Rubens under Marie de Medici, by Cal- on Periodicity. lot, and in Van Dyck's regal portraits. Holland's oblem of the relations of the several wealthy bourgeois women in black satin with white to each other, as they integrate into lace, painted by Codde, Rembrandt, and Terborch, tyle of dress, or the structural skele- are also typical of the scene. The Spanish Haps- ,style, will be touched on in still burg portraits of Velasquez illustrate a court ection, VI, on Interrelations of Dimen- dress of extraordinary uniformity and enormous d again in the interpretation attempted width. VII on Variability and Stability. After 1660, a fine series of fashion plates by Bonnard shows a narrowing skirt. As early as 1665 the farthingale is ousted from general wear, WIDTH OF SKIRT but not at court.8 For a while there is in France, as elsewhere, considerable discrepancy hries begins with a fairly wide skirt, - jed by the smooth, padded cone-shaped 6c. Kdh1er and E. von Sichart, A History of fthe Spanish fashions of the late six- Costume (New York: G. Howard Watt, 1933), 237. itury, or the squarer Queen Elizabeth Cited hereafter as K85h1er. Xowever, the year 1605 lies in the midst 7Kbhler, 314. dual tendency toward narrowing. The 8K6h1er, 288. 128 ANTHROPOLOGICAL RECORDS between court regulations and the freer flowing currents of general fashion, but in the end it is court dress that has to yield to the pressure of the narrowing trend. In the latter part of the reign of Louis XIV, beginning about 1680, the farthingale starts a long slow recovery. In Watteau, then Hogarth, and later in the magnifi- 60 50 40 30 20 10 0 10 20 30 40 50 60 cent Versailles galaxy including Boucher and I I I I I I 11111 I Nattier, we see a gradual approach to the maximum of skirt width reached about 1750. The Paris gowns that find their way to the American colo- 30- nies corroborate, though with less luxury, the fashions seen in the portraits of Mme. de Pompa- 20- dour. After 1750 or so, the trend of the eighteenth century, though still hampered by court regula- lo- tions, is a steady narrowing. A jog in the decade 1771-80 coincides with Marie Antoinette's reintroduction in 1774 of the wide flat farthin- 1900 gale, exquisitely depicted in the engravings of Moreau-le-Jeune. In general, though, what is 90_ lost in width is made up in the train. The works of Chodowiecki, Reynolds, Romney, and Gains- borough, and the fine engravings of France and 80- England of the later eighteenth century recall the general picture. By the time of the French 70 Revolution the farthingale is discarded, and a _ mere pillow at the back gives the necessary full- ness. One would have said that the imitation of 60- classic dress during and after the Directoire was a novel idea, symbolizing, perhaps, the be- _ ginning of a new and "natural" life. Instead we see that the clinging skirts are merely the cul- mination of a drift that had its inception fifty 40- years before. The years up to 1800-08 derive fullness from the trains that are occasionally found. 1810-11 is the bottom of the curve, with a trainless, short skirt. Then the trend turns slowly to rise. Its 20- peak in 1859-61 (figs. 2, 4, 6) was the result of the adoption of the farthingale again, now called crinoline. Afterward a gradual narrowing 10- set in, destined to attain its limit of possi- bility in the middle twenties of our century. 800- Since then, skirts seem to be gradually widening. A return, several decades hence, to a crinoline wider than the wearer might be difficult in this 1790- day of automobiles, but the effect could be achieved with a train. Fig. 6. Skirt width and waist width, With peaks of fullness at about 1570, 1660, sions 5 and 6, showing generally inverse 1750, 1860, and of narrowness around 1625, 1680, tion, 1787-1936. 1810, and 1925, the full period is therefore about a hundred years. Oscillations within these huge curves now seem somewhat dwarfed. Many of them are due to the temporary introduc- tion of the train, which seems to have little effect on the general drift. RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 129 LENGTH OF SKIRT toward long-waistedness sets in. Our previously established swings prove but oscillations in a dimension is also analyzed in terms of greater cycle. One hundred nineteen years after ech being the time within which the 1807, in 1926, the lowest point is reached. The a and returns to the ground. The tide has probably now turned into the second half fo maximum length are the double decade or upswing. One would have said that the long 1794, 1860, and 1902-05, with a near- waists of 1926 were another cultural abnormality in 1934-36. These suggest a diminish- following the war: they prove to be the normal length. This curve has several special trough of a tremendous curve. Moreover, one re- ,too. The 1860 and 1905 maxima are calls with interest the similarly long-waisted, by a recession that remains rather girdled frocks of 1450-1550, pictured for example d they may have to be counted together. in the early Gobelins and in the court of Maxi- r doubleness occurs in the minima. It milian I. ult to decide whether 1813 or 1833 t the true extreme of skirt brevity in DIAMETER OF THE WAIST rnineteenth century. Moreover, the th-century minimum falls around 1780, Inasmuch as it is impossible to perceive the cade and a half before the century's trend in this dimension from the 1844-1919 data ider the Directoire of 1794. Similarly, alone, Kroeber's former estimates prove to relate et minimum of 1927 is preceded by a chiefly to minor irregularities. minimum of 1927isprceIn 1605 a sturdy corset' achieves the tiny mum only eight years earlier. At this waist that for some time has been de rigueur on rt nearly reach their upper limit of every formal occasion. One may call to mind some ity, and probably our less definable of the portraits of Queen Elizabeth. This corset a decency. Glib explanations of the goes out gradually in the next few decades. orces range through such things as war- Voluminous sleeves make it hard to get waist di- gency, relaxing of sex morals, driving ameters for the seventeenth century. Our few cars, a passing tendency toward non- measurements are supported by an impression ty to emphasize equality with men. But gained from the Van Dyck and Dutch portraits that perficial consideration of the whole waists are amply wide up to 1660 or so. The dress-function and other aspects of sharp drop soon after is the result of the re- uette, reveals the inadequacy of these introduction in 167010 of the corset. After 1680 The violent increase in variability the trend changes. The considerable width for years. themselves in the late nineteen- most of the eighteenth century is surprising, be- ll be recalled, the result of the cause one has the feeling that waists are quite slender. Though court dress is undoubtedly more between the knee-length dress and the exacting in this respect, the effect of slimness oe de style." Since then the styles is achieved without real constriction partly by more homogeneous. the slenderizing V of the corsage, and partly by peculiarities of the periodic curve juxtaposition to full skirts. A minimal extreme skirt length and its variability will seems only to have been reached just before the iscussed more fully in the statistical fall of the French monarchy, after which diame- IV, V, and VII. ters increase. A loose sash tied in the back is typical of the Revolution itself. With "Greek" dress as a pattern during and after the Direc- POSITION OF THE WAIST toire, the waist is wide. Following to a certain extent anatomical exigencies, though this we well the beginning of the intensive year-by- know is not always so, waists increase in diameter zrd in 1787, there have been crests of as they shorten. The widest are those of Empire tstedness in 1807, 1869, 1917, and 1936 dresses, directly under the breasts. to be reached; and intervening maxima There now starts a very gradual constriction. Iists in 1842, 1903, and 1926. The At the turn of the trend during the eighteen- hus s n as time goe on. Thi fifties and sixties, the diameter approximates sholdaso if the more in. Teit Queen Elizabeth's famous edict (fig. 6). The data are inclued, breathing spell of 1872-75 is negligible. Not .data areincluded. .until the opening of the twentieth century does er, the extreme minima and maxima are the enlarging of the waist show itself in earnest. MLhe curve begins around 1605-10 with a The maximum comes in 1923, just one hundred and Iwaist, but sinks with the advent of sixteen years from the previous peak. tted V-waist. Its rise again to the 1807 is slow except at the very end. 9Kohler , 320 : "invented (by the Spanish) in es a period of about two hundred years. the first half of the 16th century." Julu then swings back. A slow drift ?K6hler, 320. 130 ANTHROPOLOGICAL RECORDS DECOLLETAGE WIDTH OF DECOLLETAGE It is difficult at first to see major sig- It is indeed true that "this trait nificant movements in this endlessly fluctuating to have a very long periodicity."12 Th line. The longer span of three centuries re- ruffed necks of 1605 open the series at veals a more significant form of the curve. The limit of possibility. Wide lace collar ruffed high neck of the Spanish sixteenth-cen- a more open decolletage, and the outer1 tury mode persists briefly in France. French possibility, a point several inches dow decolletage is mostly low during the seventeenth arm, is reached about 1670. A gradual and eighteenth centuries, under the particular ends in the close kerchief of the Revol in-?iuence of court dress. The fichued throat It should be noticed that the width of of the French Revolution is the apex of height, tage of eighteenth-century dress and ey then a new decline ensues. For 1844-1919, "the dress is not great. It is depth that i most striking event in the history of decolle- sized, so the general effect is that of tage depth is its increase in recent years."1l square. A rapid widening in the early This reaches its maximum around 1915. A reverse teenth century brings a peak around 183 culminates in the rather high necks of 1934-35. then for a hundred years the trend has b The evidence is not wholly satisfactory, how- toward less exposure of the shoulders. ever, on which to base a periodicity. IV PERIODICITY It seems worth inquiring in how far there may Actually the situation is often less be any more or less constant duration to the because of minor crests and troughs. S swings of the fashion pendulum just described; mension of 20, the lowest in a century, whether there is any period of years within or gradually over fifty years to a value of" near which such swings tend to accomplish them- which is the highest reached in a cent selves. There is, of course, no necessary If about halfway up the climb there is a reason why even in one feature or dimension the sion which in five or six years carries time for change from one extreme to the opposite from say 31 back to 28, after which the and back again should be constant over several march is resumed, this recession is obvi centuries, nor why the rate of change should be minor fluctuation. It is no doubt styli the same in separate features within a given and historically significant, but less so century. At the same time there might be a the longer and larger swing from 20 to 40 cause or causes tending to operate toward uni- which it is superimposed. Suppose, howev formity; and in so far as there might be, the that in twenty years the value of 35 has first step toward its recognition would be de- reached, that the recession lasts ten ye termination of the degree of uniformity which carries down to 25, and that in the foll exists, and of the time value expressing the twenty years the value mounts again and uniformity. 40. What have we then? Still a seconda On the whole, the style changes are so long tuation, although an accentuated one? Or in their range, and so progressive, that there swings of 20, 10, 20 years duration and 1 is no great difficulty, as soon as data are suf- 15 points amplitude? Obviously, if the ficiently ample, in determining recurrent maxima is to be not wholly subjective or arbitr and minima. We call a full wave length the time has to be rendered, so far as possible, interval from one crest to the next, or from of all known values for the trait in quest trough to trough. For instance, skirts were also in comparison with the trend of dur clearly at minimum width in 1811 and again in of undeniable swings; and in some degree 1926; at their fullest, not far from 1749 and reference to high and low values and dur about 1860. This gives wave lengths of 115 and in other traits. At that, there are lik 111 years respectively; or half wave lengths-- remain some cases difficult to adjudge, one-way swings--of 62, 49, and 66 years. For others dubious because of insufficient dat waist width, correspondingly alternating maxima but enough may emerge clear-cut to justify and minima fall at about 1780, 1807, 1860, inquiry. 1923,'3 giving wave lengths of 72 and 116 years, and one-way swings of 27, 45, 71 years. SKIRT LENGTH - ~~~~~~~~~~~~ ~ ~ ~~~(Dimension 2) K1lroeber, 257. The first half of the seventeenth centa E2roeber, 256. a time of full-length skirts. From 1605 t 1The moving-average figures are used. half of the dress pictures measured have RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 131 100; for only one of the years on which there moving average has fallen below 98; by 1914, be- data does the average fall below 96. From low 92; by 1919, below 85. This, the year in to 1660 full length is even more unanimous: which Kroeber's first study terminated, seemed of nine specimens available are full 100 the limit. However, by 1925 the moving average cent. We can take the midpoint of this fell below 80, and in 1927 dipped to the minimum le decade, namely 1650-51, as the moment of of 72 (year figure 69.7). of length. Once this corner was turned, the lengthening Thereafter there is slow shortening. From was rapid: 1929, 79; 1930, 84; 1931, 90; 1934, 98. to 1690, three years show 100; six years, If now we summarize, these are the maxima and es of 96 to 99. Then there is a twenty minima: I' gap in data. From 1711 to 1730 the h is still about the same as before the TABLE 9 om 1731, however, shortening increases . o essively to about 1780: successive decades Dates of Maximum and Minimum Skirt Length* e 96.6, 96.6, 95.8, 94.8, 94.2. The peak vidently comprised in 1778-81: 92.4, 92.3, Years Maximum Minimum 3, 92.2. c. 1650 .100 ow follows a rapid lengthening, expressed in 1780 90 or percentages. The 1784 value is up (skirt 1792-96 .99 *r), 1787 reaches nearly 99, 1790-9314 are 1814 .92 over 98, Directoire styles of 1794 touch 1823 95 95 The moving-average crests are 1792, 98.7, 1836 100 92 4796, 98.6. 18860.96... 0 is full length is only momentarily held, 1903-04 .100 ontrast with the long persistence around 100 1927 .72 century before. The years 1795-1803 oscil- ? 1934 ....... 98 between 97 and 100. Length decreases pro- *Moving-average values after 1788. sively from 97.6 in 1804 to 91.9 in 1814.15 i, however, not certain that 1814 is the It will be seen that two extremes have been crest of the wave, because a second peak of double peaked: the short skirt of 1814 and 1833, tness comes at 1833 with 92.0.16 Between the long one of 1860 and 1903-04. Passing over a swell of partial lengthening of the hem. the recessions within these as minor, we have: whole double decade 1814-33 is something of *1650, t1780, *1794, t1833, *1903, t1927. The it: it has shorter skirts than any period in resulting wave lengths, crest to crest and trough preceding two centuries. to trough, are 144 (1794 minus 1650), 53, 109, any rate, skirts lengthen progressively and 94 years. The average of these four is 100 1833. By 1836 the moving average has years--the empirically found value of the skirt- Ned 94; by 1841, 97; 1846, 98; 1858, 99; length cycle in modern European history. 1 99.8. The first individual year to show 100 per cent length for every dress examined 9; thereafter three more such years occur TABLE 10 dically until 1875; then no others until Summary of Periodicity in Skirt Le Between 1875 and 1899 there is a minor ening, reaching 96.1 in the moving average 1650 .... Max. 1886 and 95.5 actual mean in 1887. This 1780 .... Min. orary recession is steady from year to year, 1794 .... Max. Max.-Max.: 144 years up and down, but is modest. The real maxi- 1833 .... Min. M. -Min.: 53 years -of length comes in 1903-04, with 99.9 in the 1903 .... Max. Max. -Max.: 109 years average, and the four successive years 1927 .... Min. M. -Min.: 94 years to 1905 showing actual averages of 100. Average of four waves: 100 years * ce 1905, no year has shown a full 100. about a quinquennium, the shortening was d and barely perceptible. But by 1911, the SKIRT WIDTH 4From here on, moving-average maxima and (Dimension 5) I are used. [The lowest individual year i s 1813, with Skirt width shows three cycles in three cen- 9 ~~~~~~~~~~~turies: four peaks of slimness, three of full- ^,No data for 1833; minimum year is 1834, ness. 132 ANTHROPOLOGICAL RECORDS TABLE 11 36, or over a third. These are moving-a Dates of Naximum and Minimum Skirt Width values; the actual means for particular are somewhat more extreme: 17 for 1809 16, 38 in 1927. The several minima--or Years Maximum Minimum waist peaks--fall between 18 and 25; the 1629 42 of low-waistedness between 29 and 36. 1651-60 ...... 96 (1659, 126) These are the indicated crests: 1678-80 .39 1749-51 ...... 101 1811 *........ 27MA 1 . 1811 27 ~~~~~~~~~~~~~~TABLE 13 1861 ......... 105 (1859, 116) 1926 20 Periodicity in Waist Lenath These convert as follows: High Low Interval waist waist n.-Min. TABLE 12 1605-10 21 TABI;E 12 ~~1661-70 31 Summary of Periodicity in Skirt Width 1711-20 .... 26 108 1751-60 29 Year Extremes Intervals, 1817 ....... 18 101 years 1842 29 1629 ......... Narrow 1869 . . 22 52 1655? ....... Full 1902/03 31 1679? Narrow 50 1917 ....... 24 48 1750? . ........ Full 95 1926 36 1811 .....Narrow 132 Aeae..7 1861 .Full 111 Average ... l_l_77_|| 1926 ......... Narrow 115 The average of 77 and 65 gives 71 ye The average is a fraction over 100 years. the mean duration of a complete wave. The histories of costume give 1570 as the The decrease of wave length is stea maximum of farthingale expansion. This date lies notable. about eighty-five years before the 1655 peak of fullness. The inclusion of this earlier wave length would reduce the average from 101.3 to WAIST WIDTH 98 years. So far as the periodicity of this (Dimension 7) skirt width has changed during the last three and a half centuries, it seems to have slowed. Waist width or diameter fluctuates V about 8 and 16 per cent of figure heigh fashion plates. The duration of its s WAIST LENGTH contrary to that of waist position--doe (Dimension 3) pear to be shortening. Early data are few, on account of arm interference and, From here on, we are concerned with smaller ventions. (See table 14 on opposite p measures--waist and shoulders--and our values, The mean full wave length is 93 year which are percentages of the body height, run rather low variability from 72 to 116. lower. For our first dimension, length of waist, the periodicity is also somewhat less. It must be remembered that this measure re- fers to the vertical height of the narrowest DECOLLETAGE part of the middle of the silhouette figure. (Dimension 4) The belt, or demarcation of blouse and skirt, may fall lower, especially in front, but has For the first thirty years of our re been disregarded because it is not always pres- was no decolletage in the modern sense. ent. next thirty, it alternates with mere ne The minimum position found (lowest percentage, ing. Thereafter, we can properly speaki highest position on the body) is 18, or less exposure as standard in court or formal than a fifth of the body height;17 the maximum, However, there is a long period from ab - to 1780 in which shifts of trend are diti 17As body height has been calculated from the define, although depth of cutout grows mouth, the place of the waist line at its high- greater. Around 1784 there is a swiftd bsody d hegh at its hiSghet tposition (lowest per- decolletage as such almost disappears, centages); probably two-fifths when lowest on out for a decade. 1794-96 bring it baa the body (highest percentages). maximum is reached in 1803, with a moviS RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 133 e of 14. Twenty years later, around It must be remembered also that the dimensions there is a minimum just below 12. By of skirt and waist length and breadth express figure is back above 13;18 and for a the outline of the dress or figure as a whole, TABLE 14 Periodicity in Waist Width Extremes Extreme value Extreme Approx. Approx. -ade s (1O-yr.av.) of year year max. min. Interval i 15.8 c.1645 16 [L80 .. .. . . . 9.8 c.1675 10 1o40 . .. . . . . . 13.8 c.1735 14 90 >'80 ..... . ....11.9 1780: 8.6 c.1780 9 105 10. 14.0 1807:14.7 1807 15 72 960 ....... 8.2 1860: 7.8 1860 8 80 30 ....... 15.1 1923:15.7 1923 16 116 te___e 93 93 *,to 1931 inclusive, the moving average but decolletage really refers to a feature in- few actual year values) does not fall ternal to the figure. As regards dress, in the .5. During this century, the moving literal sense, it is negative instead of posi- several times goes above 14, with minor tive. i 1841, 1874, 1879, 1898, 1905, and a d highest one in 1915 at 15.4. The in- TABLE 15 dips in the curve--decolletage short- . . . e, however, so slight that they can- l be construed as basic waves. They Period Value Year Approx. A prix Intervals oscillations in a slowly deepening max. mn. which climaxes around 1915. From this 1605-10 5.4 max the raising of the decolletage 1631-40 .... 1.8 1635 11 values) is gradual, with even a re- 1641-50 .... 06.6 1645+ 7 to more than 14 as late as 1928. With 1771-80 .... 615.2 6 778 1 8 143 wever, the break comes: the decolletage 1788 . 8.6 1788 9 143 ferred to the back, the neck-line rises 1803 .14.2 1803 14 25 in front. A seeming minimum is reached 1823, 1827.. 11.7 1825? 12 37 *hree years: 1934, 11. 1841 . 14.6 1841 15 38 lclear that after decolletage once was 1868-69 .... 12.6 1869 13 44 (or reaccepted) as a feature of formal 1915 . 15.4 1915 15 74 a the early seventeenth century, it 1934.11.0 1934 11 65 to remain established with only minor Average... 70 72 ions, except for three sharp but tempo- ses. The first of these was in 1784-94, lletage proper simply went out. The If we consolidate the second and third waves # a longer period, marked in 1806-13, with the fourth, as there is possible warrant more so in 1820-28, with peak about for doing because their extremes are less accen- en decolletage remained in force, but tuated, we have left two crest-to-crest waves-of y sacrificed depth in order to obtain 143 and 137 years respectively, and two trough- t of breadth. The third was the recent to-trough ones of 143 and 144--a mean of 142, when the breast was covered to expose instead of 71 as in the table. These mean wave lengths do not signify too situation as regards this proportion, and much. What we can also see is two long spans, tors at work in its changes, are there- 1645-1778 and 1788-1915, in which decolletage a somewhat different order from those gradually deepens to a maximum; punctuated by in the proportions so far considered. several brief reversals, c.1635-45, 1778-88, 1803-23, 1841-69, 1915-34. This gives for the view of the tendency toward inverse re- periods of deepening of decolletage an average between depth and width of decolletage, of 140 years, for those of raising, 18 years. da in sections III and VI, it should In otedod,tewv rfl smrel be noted that the minima and maxrima of skew other uoud,th crveep slowly downward,tenl Ddimensions do not coincide in time.skw Thcuotrep sl lyd nad,hn minimum in 1823-27; width, maximum in jumps back up rapidly. 134 ANTHROPOLOGICAL RECORDS DECOLLETAGE WIDTH and then increases again for a long time. (Dimension 8) inverse relation is reflected in the ratio the two dimensions, as shown in table 18. Decolletage proper does not exist in our ratio (4/8) sinks rapidly for 30 years to series, for width any more than depth, before 31, then climbs slowly for 85 years to 1912 1630. After that date, it gets established rather more rapidly, however, and reaches a peak, with nearly full shoulder exposure or per- COMPARISON centage value of about 21, around 1659-68. Our data are blank between 1680 and 1710. After If now we bring our six sets of periodic that, the value declines gradually through the findings together, we have the following: eighteenth century, at any rate during its sec- ond half; the trend in the first half is not quite clear. The minimum width is reached with TABLE 17 a value of 8 in 1795, a few years after the vleo in15 Comparison of Six Periodicities sudden minimum of decolletage depth in 1788. C From here there is a climb of nearly forty years Dimensions Mean wave to above 23 in 1832. Thereafter the record is lengths, ye one of slow narrowing to the present. The last No. 2 Skirt length ............ 100 year of the moving average, 1934, is lower, with No. 5 Skirt width ............. 100 11.3, than any year since 1800, 11.6. The last No. 3 Waist length ............ 71 actual year, 1936, shows 9.8. The long swing No. 7 Waist width ............. 93 toward narrowing of neck-opening may thus be not No. 4 Decolletage length ...... 71 yet completed. No. 8 Decolletage width ....... 154 There are several temporary recessions in the century of drift: 1847-53, 1878-83, 1890-95, Mean of si ................ 98 1900-03, 1917-22. But these add only one to two points, each time, while the century-long de- We are not wholly clear how much weight crease drift is from 23 to 11; so they clearly be attached to this clustering of the wave are superficial oscillations. As with depth, of change in our six dress dimensions aro there is a tendency toward slow creep--in this value approximating a century. The questio instance toward narrowing; broken by definitely in how far the significance lies in the int briefer and sharper-curve reversals. fact of a century-value, or, on the other in the nearly synchronous clustering of pe certain periods, which might be due to a c TABLE 16 cause. This problem is discussed further i Periodicity in Decollqtage Width section VII. One thing, however, is certain--whether Years Value Intervals the six mean wave lengths do or do not bear lation to one another--namely, that women's, 1659-68 .... 21.2 fashions change slowly, as regards the fun 1795 ....... 8.3 proportions of the silhouette or contour. 1832 ....... 23.4 168 average, any one proportion is a half-cent 1936 ....... 11.3 141+ swinging from its extreme of length or ful to extreme of brevity or narrowness, and an half-century swinging back. This is more t It is evident that the wave profiles are would usually be supposed, in view of the e again skew. But decolletage width mounts rapid- lized world's general assumption that women ly to its maximum and then shrinks gradually, dress fashions are in their nature not only whereas decolletage depth is reduced suddenly stable but capriciously and rapidly unstabl, V OSCILLATIONS also seems worth while to try to estimate 1788 to 1934; that is, 38 periods when the ac- rage duration of minor fluctuations or tual value moves from above the moving average et oscillations over and above the major to below it and back above again (see tables 3, or trends so far considered. This would 8). This gives an apparent mean of 3.87 years, first of all the reliable determination o=4.11.19 long-time trends. The deviations from Dimensions 4, 5, 7, 8 show 40, 38, 38, 41 f the actual averages for each of a se- oscillations in the same one hundred and forty- f years would then give the periods for seven years; the mean for these dimensions is the actual style, with respect to any one 3.77 years. remained above or below its underlying Dimension 2 cannot be directly compared with these, because the measured values often reach regards statistical execution, however, 100, but cannot pass it: a dress as worn may ter is not so simple. On account of the surpass the distance to the ground, but in the number of measures, no mean for a particu- fashion plate, as soon as the toe becomes in- is very reliable; and its probable er- visible, the measurement has to be read: skirt *d have to be taken into account. More- length - 100. The range of variation is there- the basic trend can be differently ex- fore small when dresses are near the limit of for instance by three-, five-, nine-, length, and the number of oscillations would een-year moving averages; or by more rise to 55 and the mean duration fall to 2.5 te methods. A moving average from its years. If we count only fluctuations passing ture produces certain apparent fluctua- from below 99.5 to above 100.5 of the moving in the actual annual means, even when the average, or vice versa,20 the number of oscilla- are proceeding in a regular curve in tions is 40, and their average duration 3.68 a peak. Suppose for instance succes- years. ual values run 69, 70, 71, 72, 73, 75, The average length of oscillation, between 87, 82, 78, 75, 73, 72, 71, 70, 69. 3.5 and 4 years, is not far from the average sharper rise begins, the moving average duration generally assumed for the business cy- ahead of the actual values; at the peak cle. This is probably a coincidence. The value onsiderably lower; then falls more slowly will scarcely be very significant until there t it is higher for a while. In other are more individual measurements available for in the rounding of such a corner, we get each year and until more technical statistical epartures of the two lines from each consideration is given the moving-average "trend" that is, of the year-by-year graph and which forms one of the two variables whose rela- ing-average graph; though each is a per- tion expresses the oscillations. ymetrical curve. In short, three ap- The size of deviation of the actual average fluctuations are-due merely to the mathe- for each year from the moving-average trend is, properties of the technique used. however, almost certainly significant for sta- De the fewness and variability of the bility of style, as discussed below in section measurements thus appear to render an VII. te treatment hardly worth-while, we ooeed to see cursorily what the surface 19For one-way fluctuations (intervals between are. crossings Of the two lines), M= 1.95, = 2.13. sion 3 shows 38 full oscillations of soThat is, the year-to-year line not only means from the five-year moving average crosses the moving average, but crosses it with one hundred and forty-seven years from a motion of at least 1.0. .~ ~ ~ ~ ~ ~ ~ ~~~~15 VI. INTERRELATIONS OF DIMENSIONS We have computed about half the relations be- with accentuation, in the ratios. Also, a tween dimensions, and present them herewith in length varies so little, proportionately t summary. (Table 18.) The variations come out very other dimensions, that any ratio into whic much like those for diameters considered alone. enters becomes largely a function of the ol TABLE 18 Interrelations of Certain Dimensions. by Five-Year Periods. 1788-1936 Period 5/2 7/3 4/8 4/3 7/5 8/5 7/8 1788-91 ...... 54.6 50.2 105.7 37.6 23.1 16.3 141.4 1792-96 ...... 50.9 59.4 114.3 44.2 25.8 16.8 *153.6 1797-1801 45.2 67.2 *121.4 65.1 29.1 23.3 125.2 1802-06 ...... 44.4 70.7 97.8 *72.3 30.7 32.1 96.4 1807-11 ... t30.6 *72.3 88.5 63.1 *48.1 47.4 101.4 1812-16 ...... 38.1 69.8 70.9 64.6 37.7 *49.1 76.7 1817-21 ...... 43.3 56.3 83.2 64.8 27.7 38.3 72.3 1822-26 ...... 47.8 44..8 62.6 49.4 23.8 43.9 56.8 1827-31 ...... 54.0 43.4 t54.5 t45.7 23.1 44.5 51.8 1832-36 ...... 69.3 36.5 63.3 48.3 15.5 32.4 47.8 1837-41 ...... 67.3 31.9 71.9 50.2 14.0 30.6 45.7. 1842-46 ...... 60.5 30.7 73.1 49.1 14.9 33.7 45.6 1847-51 ...... 63.6 t29.6 67.0 47.4 13.6 32.5 41.9 1852-56 ...... 80.0 30.5 65.2 49.1 10.7 26.4 t40.6 1857-61 .... *103.1 31.5 73.5 52.9 t7.9 18.0 43.8 1862-66 .... 101.7 35.6 74.0 55.5 8.3 tl7.5 47.5 1867-71 .... 88.0 40.8 75.5 57.8 10.2 19.2 52.9 1872-76 .... 82.6 *43.7 87.6 *61.0 12.3 19.6 62.7 1877-81 .... 67.0 33.5 98.6 54.6 13.2 21.8 60.4 1882-86 .... 57.1 31.6 87.7 50.8 15.2 27.9 54.5 1887-91 .... 54.7 31.7 103.8 49.6 16.7 25.2 66.2 1892-96 .... 59.0 32.9 96.5 49.6 15.8 24.8 63.9 1897-1901 59.3 t29.5 106.7 48.0 14.9 22.7 65.7 1902-06 .... 5.2 31.2 111.5 t47.4 17.4 23.8 73.3 1907-11 .... 39.2 44.2 107.2 51.9 29.3 32.1 91.2 1912-16 .... 40.5 53.8 *119.7 *61.0 36.1 34.2 105.5 1917-21 .... 35.9 *54.7 113.6 55.9 45.7 41.1 111.2 1922-26 .... t26.4 45.7 104.5 40.7 *71.4 *60.8 117.4 1927-31 .... 29.6 44.3 119.5 43.7 61.1 50.4 *121.2 1932-36 30.6 47.2 97.3 44.0 39.2 37.5 104.4 This is due largely to the fact that the six measure. On the other hand, waist length, separate dimensions mostly show significant maxi- shows five crests instead of the usual t ma or minima in the decades around 1810 and 1920, since 1800, produces five crests in those so that these tend to be repeated, sometimes into which it enters. (Table 19.) TABLE 19 Periodicity of Dimension Interrelations Horizontal and Vertical Ratio of Same Part of Dress: 5/2: Minima, 1807-11, 1922-26 interval, 115 years 7/3: Maxima, 1807-11, 1872-76 interval, 65 years Maxima, 1872-76, 1917-21 interval, 45 years Minima, 1847-51, 1897-01 interval, 50 years 4/8: Maxima, 1797-01, 1912-16 interval, 115 years Vertical Diameters inter se: Horizontal Diameters inter se: 4/3: Max., 1802-06, 1872-76 (70 years) 7/5: Max., 1807-11, 1922-26 (115 Max., 1872-76, 1912-16 (40 years) 8/5: Max., 1812-16, 1922-26 (110 Mmn., 1827-31, 1902-06 (75 years) 7/8: Max., 1792-96, 1927-31 (135 [136] VII. VARIABILITY AND STABILITY OF STYLE question of when, under what circum- YEAR-TO-YEAR VARIABILITY , and why traits of fashion are rela- (Percentage Deviations of Annual Mean stable and unstable is approached by us from Moving Average) ways. is a year-by-year comparison of the The percentages by which each annual mean de- deviations of the means for each trait; viates from the five-year moving average for the B, the variability inter se of the actual same year--the basic data for this section--are ents which go into the annual average. given in table 20. More convenient are tables L probably the most satisfactory expres- 21 and 22, which express the same values averaged f stability and instability. for five- and ten-year periods respectively. We second method is to compare each annual have thought it unnecessary to diagram these re- with the "trend" or moving average for sults separately; in substance they are shown in e year. If the latter is held constant figures 1 and 2 (pp. 114,115), where the line how many per cent above or below 100 is represents the moving average, and dots the ual average for the year? Thus for skirt annual means. the moving average for 1801 is 45.2, the mean 42.1, or 6.9 per cent less. For TABLE 20 on the other hand, the moving average has only to 45.6, but the year's mean is Percentage Deviations of Actual Year Means from or 31 per cent higher. For 1803 the devi- _Trend, 1_88_1934 is 10.3 per cent under. Obviously this Year 2 3 5 7 iod in which the style for skirt width y variable from year to year, even 1788 .. 1.4 14.0 3.5 2.0 32.0 the trend is pretty consistently in one 1789 .. .8 5.3 12.0 3.4 21.1 23.0 on for two decades. By contrast, the 4-58, which also show a strong one-way 1791 .. .2 1 19.4 5. 23 16.8 4.6 n this dimension, run 101.1, 101.7, 101.8, 1792 .. .2 6.7 8.9 13.2 12.3 100.3; and 1839-43, with the trend change 1793 3 4.4 47.0 .4 4.5 8how 97.8, 99.1, 103.5, 98.2, 99.8. It 1794 1.7 13.4 3.1 13.4 11.6 15.5 n that the year-to-year fluctuation was 1795 1.6 3.4 25.2 9.1 1 2.1 14.6 re marked in 1801-03 than in 1839-43 or 1796 .4 3.0 12.1 6.2 28.6 . In other words the fashion, with re- 1797 1.2 3.0 11.8 5.8 11.2 8.0 to this trait at least, was much less 1798 .2 1.0 .9 .4 21.1 4.0 in its trend in the earlier period than 1799 .3 1.0 7.2 3.6 13.2 7.8 two later. 1800 .. .1 4.2 19.8 7.5 1.6 1.7 objection which can be made against this 1801 . 1.6 4.8 16.5 6.6 26.0 2.5 measure is that it eipresses the rela- 1802 .. 2.6 3.2 21.6 30.7 13.7 12.9 an actual year average to a short moving 1803 .. .4 1.1 11.3 12.5 6.6 5.1 to which it contributes; also that the 1804 .. 2.6 3.2 9.6 3.2 1.5 2.9 average, our base, possesses properties, 1805 .. .4 1.6 14.3 10.0 5.0 21.2 tion to the actual sequence of events, 1806 .. 1.3 3.8 .8 2.1 .7 8.6 vay according to the nature of the se- 1807 .. .1 .5 10.6 12.2 3.4 5.8 of events. It behaves somewhat differ- 1808 .. .6 1.1 2.5 3.2 1.4 2.2 hen it is steadily progressing in one 1809 .. .2 10.8 4.9 15.4 6.4 1.4 on and when it is turning a corner; and 1810 .. .8 5.0 6.3 12.3 6.5 14.6 indifferent parts of its curve around 1811 .. 1.2 8.5 6.4 1.1 2.2 3.8 It is for this reason that the series 1812 .. .3 2.0 1.6 4.3 .0 4.3 ple percentaged standard deviations, or 1813 .. 4.8 3.0 11.2 11.1 6.7 17.9 lity coefficients, is probably sounder. 1814 .. 2.5 4.8 8.2 10.6 3.0 .0 ,these coefficients directly express 1815 .. .7 7.5 7.8 21.5 2.3 3.2 .variability or instability within one 1816 .. 2.0 9.5 7.7 .3 4.8 10.7 nmuch the several fashion plates for 1817 ... 2.3 11.0 26.9 3.1 11.6 35.1 ffer from one another, for example; in- 1818 |o 1.5 8.4 14.4 3.3 .9 7.9 ity over several years must be inferred by 19 . . . . . . son. The annual deviations from the mov- 1820 1.1 8.1 10.0 7.3 6.5 1.8 rage express variability within a span 1821 2.7 4.3 17.2 2.2 7.7 2.3 s8 directly. We therefore use this measure 1822 . 0*n the whole, the two measures give re- 1823 ... 1.7 4.7 3.4 5.5 6.4 4.7 fairly in agreement. Those by the method 1824 ... 2.6 10.8 16.0 16.1 14.8 4.7 iation from the trend will be presented 1825 ... 1.1 1.6 9.2 3.6 7.3 13.6 1826 ... 1.9 .8 10.0 1.9 3.7 16.7 [1371 138 ANTHROPOLOGICAL RECORDS TABLE 20 (Continued) ____TABLE 20 (Concluded) Year 2 3 4 5 7 8 Year 2 3 4 5 7 1827 .. .7 7.2 4.3 4.7 5.7 10.0 1882 .. .7 2.6 6.6 1.4 7. 1828 ..3.0 1.1 2.5 2.9 9.1 .9 1883 .. .0 2.3 4.5 .9 2.4 1829 .. .6 .0 .0 10.5 2.6 4.5 1884 .. .1 1.5 3.0 5.3 2.4 1885 .. .7 2.2 3.7 3.5 2.4 1830 ..1.1 6.1 .8 3.6 4.3 9.8 1886 .. .3 .7 8.0 3.5 4.1 1831 ..1.6 9.5 2.4 12.8 12.5 7.0 1887 .. .6 .7 7.2 6.8 5.7' 1832. 1.5 .4 .8 2.8 5.2 1.7 1888. .5 .0 1.4 8.2 4.6 1833 ..1889 .. .2 .4 .7 2.5 9.1 1834 ..3.1 .4 6.9 .6 1.0 15.6 1835 . .6 2.6 4.5 1.7 .0 12.8 1836 .1'.8 2.5 3.7 4.6 .0 8.7 1890 .. .4 .4 1.4 5.1 5.6 1837. .0 2.5 1.5 6.7 2.2 1.1 1891. .2 1.1 4.3 2."7 0.0 1838 .. .5 3.2 .7 7.3 1.1 1.5 1892 .. .4 2.1 3.6 3.8 1. 1839. .4 .7 3.5 2.3 4.4 4.0 1893. .7 3.9 .2.2 .4 2.2 1894. .3 2.9 .7 4.5 .0 1840 .. .0 1.7 1.4 1.1 1.1 4.6 1895 .. .3 2.1 .7 1.3 5-.5 1841 .. .1 .7 1.4 3.5 3.3 3:0 1896 .. .0 2.1 .0 14.8 9.1 1842 .. .4 1.0 .0 2.3 1.1 8.2 1897 .. .4 .7 .0 2.4 2.3 1843. .1 .7 .0 .7 2.2 6.7 1898. .2 .7 .7 11.4 9.01 1844 .. .6 .3 2.1 1.4 2.3 :5 1899 .. .3 .3 2.1 10.5 5.7 1845 .. .4 2.4 .7 1.0 3.4 1.0 1846 .. .2 .0 6.4 4.3 3.5 4.1 1900 .. .5 1.3 7.9 10.9 4.4 1847 .. .4 2.1 8.0 6.6 3.5 .0 10 . . 7 1. 10 . 1848. .1 2.5 .7 3.4 .0 1.0 19021. .2 3.5 61.4 4.1 4.2 1849 .. .3 .0 2.2 .3 1.2 1.5 19032. .1 4.5 9.4 41.4.0 .1904 .. .1 4.9 4.1 2.5 3.1 1850 .. .2 1.1 5.9 .9 1.2 .0 1905 .. .2 .3 .7 .4 4.2 1851. .7 3.9 3.7 6.7 2.4 1.4 1906. .1 .7 15.7 5.1 3.1 1852. .4 3.2 4.4 1.7 .0 1.9 1907. .0 2.2 5.7 3.0 6.? 1853. .0 .4 5.9 3.6 7.1 .5 1908. .2 3.4 15.2 7.7 .0 1854. .1 1.8 4.4 1.1 .0 .5 199. .4 58 .0 .3 23 1855 .. .0 1.8 1.5 1.7 7.1 3.4 10 4 58 90 13 1. 1856 .. 2 1.8 4.3 1.8 2.4 3.5 1857. .5 .7 .7 9.2 3.6 1.0 1910. .2 .4 .7 3.8 3.3 1858 .. .4 1.9 10.1 .6 1.2 .0 1911 ..1.0 4.0 .0 25.4 5.5 1859 .. .4 1.6 5.9 12.6 1.3 1.1 1912 ..2.3 4.0 5.0 6.5 2.3 1913 ..2.0 1.6 4.8 5.6 2.3 1860 .. .0 1.6 8.9- 2.3 2.6 .5 1914 .. .2 1.6 5.3 21.6 3.7 1861. .4 .4 6.1 .6 .0 .0 1915. 1.7 2.0 5.2 7.0 2.2 1862 .. .1 1.6 2.3 5.6 6.2 .6 1916 ..4.3 1.6 8.6 21.2 5.3 1863 .. .8 3.3 .8 .5 8.4 3.4 1917 ..1.7 .4 2.0 39.7 1.5 1864. .0 1.3 3.1 1.1 1.2 2.3 1918 .. .6 2.4 7.4 42.7 .0 1865. .7 .9 .8 6.7 1.2 .0 1919. .5 5.1 .0 7.8 5.0 1866. .:6 .4 .8 .7 3.5 4.0 1867 ..15 3.6 7.9 2.6 5.8 1.2 190. 24 26 64 3.4 7 1868 .. .3 .5 .8 4.0 3.4 5.8 19210. 2.5 2.1 64.314. 1.3 1869 .. 1.0 .0 9.5 1.8 1.1 4.2 1922 2.5. 6.4 3.63 16.8 6.5 1870 .. 2 .0 9.8 6.2 2.1 10.8 1923 . 7.3 6.6 .0 5.1 1.9 1871 .. .1 3.1 3.7 8.9 7.2 4.6 1924 ..4.0 12.1 5.8 1.8 2.6 1872 .. .1 .4 10.3 5.4 12.2 .6 12 25 20 67 1. . 1873 .. .2 7.5 1.4 5.7 1.0 9.6 1926 ..3.6 11.1 13.7 16.7 4.1 1874 .. .2 3.9 .0 2.9 5.9 6.8 1927 ..3.2 8.9 6.5 13.0 3.5 1875 .. .7 4.3 2.2 3.5 8.3 9.7 1928 ..5.9 8.7 4.9 1.4 1.4 1876 .. .0 1.3 3.6 7.2 1.1 8:8 1929 ..9.9 .6 7.8 11.1 6.3 1877 .. .4 1.2 2.2 2.4 6.5 7.4 1878 .. .1 1.6 3.6 2.3 .0 2.8 1930 ..5.3 5.1 5.8 3.6 20.4 1879. .1 .4 6.3 6.2 1.1 4.2 1931. 5.4 2.2 14.6 .4 3.0 1932 ..1.6 5.8 7.3 .8 18.5 1880 .. .6 4.5 10.0 11.0 2.3 4.1 1933 .. .7 1.2 3.4 9.1 8.3 188 Q . .0 3.4 3.6 1 1.1 .0 7 3. 193 .r . a9 .0 15. 13.0Iz 17. RICHARDSON AND KROEBER: TBREE CENTURIES OF WOMEN'S DRESS FASHIONS 139 t of all, it is clear that the propor- and width (Nos. 2 and 5) attain their greatest amount of deviation varies among the six variability in post-World War years; the waist ons dealt with. On the whole, the large and decolletage dimensions vary most before 1821 rs have low variabilities. Thus, skirt --in fact, all except waist length (No. 3), be- , the absolutely largest dimension, has fore 1803. These dozen early years 1788-1799, its highest five-year-mean percentage in the moving-average record, are only fairly ion (table 21), while 19 out of 30 values conspicuous for number of maxima or minima at- tained, in spite of their high deviations. ?802- 16 show more wave crests, but the year-to-year TABLE 21 variability averages lower in four of the six ye-Year Averages of Annual Deviations dimensions (tables 20, 21, 22). * from Trend, 1788-1934 d 2 3 4 5 7 8 TABLE22 . .8 7.3 16.8 3.0 21.5 14.0 Ten-Year Averages of Annual Deviations 6 .8 6.2 19.3 8.5 13.8 15.0 from Trend, 1788-1934 801 .. .7 2.8 11.2 4.8 14.6 4.8 6 1.5 2.6 11.5 11.7 5.5 10.1 Period 2 3 4 5 7 8 .0 .6 5.2 6.1 8.8 4.0 5.5 1788-96 . 8 6.8 18.1 5.8 17.6 14.5 1797-1806 1.1 2.7 11.4 8.3 10.1 7.5 . 2.215. 147.239.6 3.4 7.2 1807-16 ... . 1.4 5.3 6.7 9.2 3.7 6.4 1. 2.2 10.8 14.2 3.5 7.3 10.59 1817-26 .... 2.0 7.7 11.9 5.2 7.7 10.2 e... 1.8 4.5 9.0 6.8 68.1 9. 1827-36 .... 1.6 3.2 3.0 4.7 4.2 8.1 1.4 4.8 2.0 6.9 6.8 6.4 1837-46 . 3 1.4 1.8 3.1 2.5 3.5 6 ... 1.8 1.5 .4.0 2.4 1.6 9.7 1847-56 .2 1.9 4.1 2.8 2.5 1.4 ....2 1.8 11.7 14.2 2.4 2.8 1857-66. .4 1.4 4.0 4.0 2.9 1.3 .32 . 1.8 1.79. 2.542.8 1867-76. .4 2.5 4.9 4.8 4.8 6.2 s . 3 .9 1.8 1.9 2 .5 4.1 1877-86 . .3 2.1 5.2 4.8 2.9 4.6 1... 3i 1.8 4.1 2 3.6 .7 . 1887-96 . .4 1.6 2.2 5.1 4.3 3.5 6 .. 1 3 1.8 4.1 720 3.0 2. 1897-1906 o.2 1.8 5.9 7.0 3.6 10.9 1. 3. 1.2 6.3 5.1 1.7 * 1907-16 .... 1.3 2.7 6.0 10.3 4.4 6.9 -o4 6 5 lo6 1 2.9 1 4.1 2.1 1917-26 _ 2.6 5.1 5.0 17.7 2.8 4.1 .4 1.5 1.6 2.9 4.1 2.1 1927-34 o.j3.5 3.7 18..3 6-.8 10.O8 7..4 6 . 2 3.5 3.5 4.9 5.7 7.1 1 . .2 2.2 5.1 6.6 2.0 4.5 6 . .4 1.9 5.2 2.9 3.8 4.6 Figure 7 shows graphically all deviations of the year from the trend, above a certain magni- 1 .. .4 .5 3.0 5.1 5.0 2.8 tude. This magnitude has been chosen so that 6 .. 3 2.6 1.4 5.0 3.6 4.2 the number of large deviations represented would 01 o .3 .7 4.8 4 9.2 4.3 9.1 be about the same for each of the six dimensions. .* 41 2.8 763 4.7 2.9 12.6 Convenient values are 3 per cent for No. 2, skirt 1D 0 .0 ..4 3.2 6.1 8.2 5.6 3.9 length; 6 per cent for No. 3, waist length; and 6 ... 2.1 2.2 5.8 12.4 3.2 9.9 12 per cent for the others. These are desig- 1*... 1.5 2.5 4.0 25.1 1.7 5.1 nated in the figure as "fluctuation units." For 6 ID 3.6 7.6 6.0 10.1 3.8 3.1 instance, for trait No. 4, decolletage depth, the 1s. 5.9 5.1 7.9 5.9 6.9 4.9 year-from-trend percentage deviations beginning 4i.. 1.1 2.3 8.7 7.6 14.6 9.8 in 1788 (table 20) are 3.5, 12.0, 19.4, 32.2, 8.9, 47.0. In terms of 12-per-cent units, these equal 0, 1, 1, 2, 0, 3; and they are entered by *er 1.0. Waist length rises to a maximum as many crosses on the vertical line denoting 8, and only thrice falls below 1.0. Decol- dimension No. 4 in the figure. depth, on the other hand, rises as high The number of crosses in this figure is ap- 3, and never goes below 1.4. The trans- proximately the same for the six traits. Thus: diameters, which of course average lower he longitudinal, run about like decolle- Number of Number of years spth. .............. fluctua- in which these it is clear that while 1835-1910 is a tion units occur f small deviations or high year-to-year Dimension 2 ... 16 13 ity for all six traits, these traits vary Dimension 3 ... 28 23 erably among themselves as to whether Dimension 4 ........ 24 19 greatest instability falls in the period Dimension 5 ........ 29 22 or after the long quiet span, and whether Dimension 7 ........ 21 - 18 or late in 1788-1835. Thus, skirt length Dimension 8 . 23 22 140 iANTHROPOLOGICAL RECORDS That there is a relation between large year- to-year fluctuations and wave crests or troughs is clear from figure 7, as compared with figures 3-6, 8-11. It is also clear that the relation is by no means simple or complete. Sometimes the fluctuations pile up in the years surround- ing a peak; thus, dimension 2, 1926-27.21 In A other cases, the fluctuations are most extreme some years before or after: dimension 3, 1807; 4, 1804; 5, 1926; 7, 1811, 1923. Several times DIMENSION 2 3 4 5 the fluctuations cluster continuously between a FLUCT. UNITS 3 6 12 12 near-by crest and trough: dimension 4, 1788-1803; 5, 1912-26; 8, 1795-1832. On the other hand, * there are crests without any accompanying marked 1790 + + annual fluctuations: dimension 3, 1860, 1903; 4, * + 1892, 1902/03; 7, 1852; and smaller clusters of 1800- fluctuations remote from any peak: dimension 7, 1824-31; 8, 1897-1903. + + Essentially, each larger fluctuation repre- 10- + sents a one-year reversal of the current five- year trend. Periods of many acoentuated fluc- 20- + tuations therefore are periods in which style is as it were two-minded or under strain; even + though it may be moving rapidly in a certain di- 30- rection, the movement is meeting with resistance. + Periods of only minor fluctuation, on the con- trary, may be construed as times in which style 40- is progressing harmoniously and whole-mindedly, whether the change be rapid or slow. It is 50. clear that 1840-1900 was such a period of har- mony, although it attained maxima in fullness of skirt and slenderness of waist and near-maxima 60- in length of skirt and both long and high waist- edness. Table 23 summarizes these differences by both five- and fifteen-year intervals. 70- It will be seen that the pre-1836 period of unsettlement is really double. The fluctuations are most marked and most numerous before 1800, then diminish, to resuma again after 1815 and straggle along until about 1835. In historical 90- terms, the Revolution-Directoire epoch was highly unstable, the Empire fairly settled, the twenty years after Moscow and Waterloo unsettled again. 1900 By 1830 quiet was impending, and 1848 was well within a long calm. Unsettlement began again, in one feature, + about 1900; became acute in another in 1911; in still others about 1920, 1923, 1930. By 1933 20- * it had definitely diminished, except possibly in + one feature: waist width. It is evident that L 0 + the beginning of the era is pre-World War, its 30 peak post-War. Only in one trait, skirt diame- + ter, do the greatest fluctuations occur during Fig. 7. Frequency of deviatio the War itself. The specific cause of this ex- five-year moving-average trend b ception seems to be a sharp reversal about 1915 tuation units per year, 1788-1934. in a narrowing trend which had come to a pre- liminary peak in 1912, but did not reach its ex- treme until 1926. This extreme was reached and passed with much less wobbling. 21This is probably at least in part a function of the moving average rounding a sharp cre st . RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 141 compare the two eras of frequent annual caution must be noted. Before 1834, the average , it is apparent that the earlier, number of observations per year is well under ,is more accentuated and may prove to ten; since 1920, above ten. The annual means longer; at any rate if the quieting are therefore less well founded for the early e 1932 continues after 1936. Fluctua- era. Where observations number only five, three, all waist and decolletage dimensions two, or one for a year, fluctuations from the itely more marked during the earlier trend may be due to smallness of the random era; in skirt proportions, during the sample used; in other words, they may be appar- is difference is of interest because ent rather than real. However, the long nineteenth-century or Vic- torian calm of small fluctuations is clearly be- TABLE 23 yond possibility of doubt. Kcy of Fluctuation Units, 1788-1934 VARIABILITY WITHIN THE YEAR c,cIn 5-year In 15-year periods periods This is the standard deviation or sigma of 147 the individual measures around their mean for 17 J 37 (14 years)* the year. For uniformity among the six dimen- 6 sions, these sigmas are expressed in percentages l ... 8 19 of their means; that is, they have been converted 6 . . 19 into Coefficients of Variability, V = 100 a/M. ' 13 The full list of V's is given in table 24; E 6 1 25 their five- and ten-year averages in tables 25 ....... 6) and 26. 0X TABLE 24 0 . .. U Percentage Sigmas of Annual Means, 1787-1936 ....... ... O f 1 (V = 100 a/M) 2 ) 2 Year 2 3 4 5 7 8 ...... .. 1787 1.0 3.4 18.3 14.3 ...... .. O 1 1788 3.3 2.3 38.2 11.9 ...... 1789 3.3 7.4 34.7 10.6 25.9 56.0 i .. 2 8 1790 0.9 16.7 65.7 10.6 16.8 1791 0.7 5.8 11.2 7.1 8.3 23.3 ''. '' 4 .... 1792 1.8 3.8 ... 4 l 16 1793 0.3 6.2 59.2 4.1 16.3 1794 0.0 6.4 7.1 10.2 14.6 21.1 1795 3.3 7.6 24.6 15.8 6.9 10.7 12 29 (13 ears)t 1796 . 0.5 8.7 44.7 3.8 12 29 Y13 yearst 1797. 0.9 3.6 0.0 16.2 1.7 5 1798 0.8 5.2 12.9 7.2 11.1 22.2 te of 40 in 15 years. 1799 1.1 8.1 21.0 14.5 15.3 26.3 te of 33 in 15 years. 1800 0.7 12.4 22.1 14.4 16.1 12.9 eonic period also attained sharp cli- 1801 1.4 11.8 9.5 29.2 9.4 5.9 ,shortness and narrowness of skirt; but 1802 0.0 6.7 3.5 cefully, so to speak, as compared with 1803 1.3 11.1 17.5 39.1 7.6 16.6 27 climaxes. It would seem as if 1811- 1804 * 3.6 9.3 36.6 27.9 5.9 18.6 ated the skirt so far as it could 1805 . 1.1 8.0 8.3 24.7 12.7 12.3 ically questioning its nature, where- 1806 1.1 7.4 18.5 33.6 11.4 21.6 1807 ..1.9 10.0 18.0 28.8 10.6 10.8 , was calling its very existence into 1808 2.0 11.2 11.8 41.7 10.4 28.9 4N!temporarily trying to rupture the 1809 2.5 11.9 16.5 38.0 10.7 19.3 ern of skirt, so to speak. The earlier mewhat similarly, though on the whole 1810 3.1 12.6 13.0 37.6 11.7 18.2 ly, disturbed about waist and decol- 1811 4.0 14.9 11.3 41.1 6.6 12.6 tportions. In brief, its revolution- 1812 ..1.7 13.1 7.0 49.6 11.9 12.0 npts concerned the bust; the recent 1813 ..9.0 18.6 17.7 42.2 9.5 16.7 legs. 1814 ..5.5 8.5 4.2 14.3 8.6 15.8 ection with the somewhat greater fre- 1815 2.9 9.6 7 314 33| 971 417 17.6 -early fluctuations, a statistical 11 .28 94 3. 71 41 1. 142 ANTHROPOLOGICAL RECORDS TABLE 24 (Continued) TABLE 24 (Concluded) Year 2 3 4 5 7 8 Year 2 3 4 5 7 1817 .. 2.4 5.0 30.3 14.1 19.8 1877 .. 1.0 4.2 8.3 13.5 10.8 1818 .. 0.5 24.7 1.8 9.0 4.9 6.1 1878 .. 1.6 5.9 11.8 15.1 6.5 1819 1879 .. 1.1 3.6 10.0 13.0 4.9 1820 .. 5.7 7.5 1.9 4.6 5.8 2.3 1880 .. 1.1 3.3 8.9 25.1 4.1 1821 .. 2.3 7.9 6.9 3.8 4.0 10.0 1881 .. 1.7 4.9 10.3 30.9 5.9 1822 1882 .. 2.4 3.8 11.1 39.4 4.7 1823 .. 1.7 7.2 1.4 1.9 4.4 2.7 1883 .. 0.9 6.4 7.1 34.3 6.1 1824.. 6.5 12.6 18.1 33.0 8.9 2.2 1884 .. 1.8 4.5 6.8 25.9 9.9 1825 1885 .. 1.9 5.3 8.5 33.5 9.6 1826 . . 2.1 23.7 0.0 6.2 3.5 12.6 1886 .. 1.4 7.5 12.0 27.0 16.5 1827 .. 3.9 9.9 9.5 7.4 17.1 5.2 1887 .. 1.7 4.3 7.4 23.6 9.1 1828 .. 2.9 7.0 8.0 2.8 8.1 10.1 1888 .. 1.0 5.9 12.3 32.4 12.0 1829.. 2.5 7.9 11.6 6.7 5.4 1889 .. 2.0 5.6 8.1 26.7 13.8 1830 .. 2.3 8.2 10.4 6.2 14.9 2.0 1890 .. 1.1 5.2 9.6 21.6 7.9 1831 .. 2.5 13.9 3.4 9.6 7.1 1891 .. 1.2 6.9 10.4 27.2 7.0 1832 .. 1.2 4.5 2.2 12.6 5.5 5.0 1892 .. 0.9 6.4 11.8 22.2 7.7 1833 1893 .. 1.2 7.8 15.2 27.5 14.2 1834 .. 4.5 4.2 13.8 6.8 5.7 12.0 1894 .. 1.0 6.9 7.5 15.3 8.2 1835 .. 1.4 5.7 19.7 12.0 1.1 37.2 1895 .. 1.2 8.5 10.6 15.6 8.5 1836.. 1.7 7.5 14.2 4.6 4.1 12.5 1896 .. 1.2 5.4 12.5 7.1 13.5 1837 .. 1.4 6.9 16.2 16.1 8.5 22.2 1897 .. 0.3 3.3 15.6 9.2 5.8', 1838 .. 0.9 5.8 12.7 5.5 7.6 12.5 1898 .. 0.6 4.8 24.7 9.0 10.1 1839 .. 0.5 4.4 12.4 11.2 15.8 6.5 1899 .. 0.0 3.7 8.1 11.0 12.7 1840 .. 2.0 6.5 14.2 10.3 16.7 15.4 1900 .. 0.8 5.0 12.8 8.2 7.3 1841 .. 0.7 4.1 11.2 8.5 8.4 15.9 1901 .. 0.6 7.1 14.0 6.0 9.4 1842 .. 2.9 5.8 12.4 13.6 16.8 21.4 1902 .. 0.0 9.1 15.5 14.2 7.9 1843 .. 1.4 5.1 12.4 9.0 8.8 10.7 1903 .. 0.0 8.3 14.0 18.5 7.1 1844 .. 1.2 4.2 11.1 6.2 6.8 5.8 1904 .. 0.0 5.0 17.8 10.3 6.8 1845 .. 0.5 3.7 9.1 4.7 5.8 4.2 1905 .. 0.0 3.0 9.3 7.7 3.6 1846 .. 1.1 3.5 8.6 1.6 5.7 5.0 1906 .. 0.9 8.3 13.7 5.0 10.5. 1847 .. 0.8 5.8 10.4 1.7 4.2 6.9 1907 .. 0.8 6.2 13.2 20.2 16.5 1848 .. 0.0 4.5 10.6 3.9 5.9 7.6 1908 .. 1.0 6.2 23.1 11.3 13.8 1849 .. 1.0 2.2 8.7 6.7 8.2 6.5 1909 .. 0.6 7.4 25.1 15.7 13.5, 1850 .. 0.4 4.2 8.0 5.7 8.1 6.1 1910 1.8 6.8 17.9 24.7 7.8 1851 .. 0.9 3.8 10.9 8.5 6.0 5.2 1911 .. 1.1 4.3 15.9 24.4 7.8 1852 .. 1.3 3.7 9.6 7.4 5.5 10.4 1912 .. 2.1 8.4 8.7 19.4 9.3 1853 .. 0.7 2.8 10.1 5.4 5.6 11.9 1913 .. 4.4 7.3 20.5 37.1 15.0- 1854 .. 0.3 2.9 9.3 12.7 8.2 7.6 1914 .. 2.2 6.1 18.8 46.0 14.5 1855 .. 1.0 5.9 9.0 9.6 7.9 8.0 1915 .. 3.4 13.1 12.0 51.5 10.8 1856 .. 1.3 6.1 15.1 3.5 5.7 5.8 1916 .. 4.3 8.0 21.7 27.3 12.4 1857 .. 0.8 6.5 13.9 3.7 14.8 13.7 1917 .. 5.3 5.7 5.3 36.0 19.1 1858 .. 0.8 2.6 15.5 7.2 8.1 7.4 1918 .. 1.9 7.6 14.9 40.8 7.6 1859 .. 0.0 3.2 14.9 4.1 10.1 19.2 1919 .. 6.6 10.6 13.9 64.5 8.6 1860 .. 0.6 5.9 2.7 2.7 6.4 7.6 1920 .. 3.8 6.2 16.4 44.3 9.3 1861 .. 0.0 4.2 11.5 7.0 12.5 4.2 1921 .. 5.1 14.8 27.5 35.8 13.5 1862 .. 0.8 6.3 14.6 6.2 11.4 7.6 1922 .. 3.8 10.7 31.9 54.5 10.2 1863 .. 1.4 1.4 7.6 4.7 0.0 9.9 1923 .. 3.0 17.6 30.6 52.6 6.9 1864 .. 1.0 4.7 3.7 3.9 7.4 9.6 1924 .. 4.6 13.3 19.7 44.7 11.4 1865 .. 0.4 5.0 12.7 4.7 5.8 8.6 1925 .. 3.3 19.8 28.0 30.4 11.3 1866 .. 0.6 4.9 13.3 1.5 9.5 7.8 1926 .. 3.1 19.2 18.5 17.6 8.3 1867 .. 1.5 4.6 15.9 7.2 7.9 12.0 1927 .. 1.9 13.2 17.3 19.0 13.8 1868 .. 1.5 6.8 8.1 15.1 12.5 12.9 1928 .. 5.4 7.7 18.3 52.3 11.3 1869 .. 0.0 2.8 10.4 10.7 0.0 9.4 1929 .. 4.4 10.8 12.7 23.0 24.2 1870 ., 1.3 6.3 10.9 16.3 17.2 17.0 1930 .. 6.4 12.1 10.4 19.2 31.0 1871 .. 0.9 7.7 12.5 21.8 0.0 17.5 1931 .. 3.8 11.3 21.7 19.7 17.4 1872 ... 1.1 8.7 16.1 12.6 7.5 7.4 1932 ... 2.6 12.4 22.3 29.6 17.3 1873 ... 1.3 5.8 9.6 9.5 8.2 22.1 1933 ... 3.0 8.8 28.6 27.3 26.2. 1874 ... 1.0 3.9 4.5 12.4 5.3 14.1 1934 ... 1.4 8.1 44.7 23.8 16.4, 1875 ... 0.0 7.3 8.0 10.2 4.7 3.7 1935 ... 2.1 14.1 33.8 29.2 23.5r 1876 ..1 1.2 7.9 110.4 8.9 111.8 25.9 .1936 .. 2.1 1 0.0 32.8 39.3 117.33 RICHARDSON AND) KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 143 TABLE 25 ~~~~~It is at once evident that the variability is -YearAver es f Pecent e S as,markedly different among the six dimensions. 1787-1938ge ofPretg imsDress length, dimension 2, again shows much the 1787-1936 ~~~lowest variability, and waist length next least. (v = ioo a/____ The four other dimensions run about alike; though 2 3 4 5 7 g the two decolletage measures show a strong pre- pon'derance of V's between 10 and 20. The little *001.8 7.1 33.6 10.9 10.2 19.8 subjoined table (no. 27, based on table 25) shows 0.8 5.8 27.1 6.8 8.3 15.9 the distribution of size of five-year averaged 1.1.0 8.2 13.1 16.3 10.7 13.5 V's. 1.4 8.5 16.2 25.1 8.2 13.8 2.7 12.1 14.1 37.4 10.0 18.0 4.4 11.8 13.6 31.4 8.8 14.9 TBE2 2.6 10.5 4.9 10.3 4.2 4.4 Distribution of Size of Variability Coefficients 2.8 9.4 6.3 7.5 10.8 4.5 Fie-Year Avera -es Aong the SxMaue 1.1 5.5 13.3 10.3 11.4 14.5 Variability 2 3 4 5 7 8 1.4 4.5 10.7 7.0` 8.8 9.4 coef fi cie nt s 2__ 3__ 4__ ___ 7 0.6 4.1 9.7 5.3 6.5 6.5 __ ..0.9 4.3 10.6 7.7 6.6 8.7 1.9 or less 19 0.4 4.5 11.7 4.9 10.4 10.4 2.0-4.9 ....11 7 1 2 3 2 0.8 4.5 10.4 4.2 6.8 8.7 5.0-9.9 ....-- 17 7 8 14 5 1.0 5.6 11.6 14.2 7.3 13.8 10.0-19.9 ... 6 18 10 12 22 0.00 09 6.7 9.7 10.7 7.5 14.6 20.0-50.0 - -- 4 10 1 1 1.3 4.4 9.9 19.5 6.4 11.8 __ 1.4 5.6 9.6 26.3 10.0 12.6 It can be concluded from this that dress 1 .0.5 4.8 15.0 8.7 9.1 18.9 length, and next to it waist length, can be 0.2 6.7 14.1 11.1 7.2 18.5 varied least from the ideal norm of a given mo- p..1.3 6.2 19.0 19.3 11.9 15.3 ment if a dress is to be within fashion. With 3.5 8.6 16.3 36.3 12.4 13.9 respect to decolletage and all transverse dimen- o 4.,5 9.0 15.6 44.3 11.6 15.7 sions, the style is much less strict, and much 3.6 16.1 25.7 40.0 9.6 15.0 more variability is exercised, within the year 4.4 11.0 16.1 26.6 19.5 16.9 and within a five-year period. What our aesthetic 2.2 10.7 32.4 29.8 20.1 29.4 taste assumes as primary in the style norm, and inhibits too great departures therefrom, is the length of the dress as a whole; next, the position of the waist constriction. Skirt fullness, waist diameter, and length and breadth of decolletage are allowed much more individual variation from AveraRes of Percentage Sigmas, 1787-1936 dress to dress.2 (V = 100 O/X) ~~The first thing that is evident from tables 25 _________- ___________ and 26 is that there are once more an early period 2 3 4 5 7 8 of high variability, a middle one of low, and a recent one that is high again. A table could be 1.2 8.4 14.74 20.7 9.5 13?.7 constructed that would be similar to table 23. 3.6 12.0 13.9 34.4 9.4 16.5 Instead, in table 28 we give the maxima of V in 2.4 10.0 6.6 8.3 3.6 6.0 five-year means. ..1.3 5.0 12.0 8.7 ~10.1 12.0 22 These different behaviors of the six dimen- 0.8 4.2 10.2 6.5 6.5 7.6 sions are perhaps partly a function of their 0.6 4.5 11.1 4.6 8.6 9.6 absolute size: No. 2 is of course by far the 0.09 6.2 10.7 12.5 7.4 14.2 largest measurement, and Nos. 4, 7, and 8 the 1.5 5.0 9.5 25.8 7. 12.2 smallest. With the small dimensions, the proba- 1.1 6.3 10.6 21.9 10.2 14.6 bility of error of caliper measurement is 0.3 5.8 14.6 9.9 8.2 18,7 greater, presuimably tending to increase thie 2.5 7.4 17.7 27.8 12.2 14.6 variability. However, the differences are not A...*4.1 - *12.6 - 2.7 *42.2 1.6_ 1.4_ wholly a fuinction of siz,e, because skirt width 144 ANTHROPOLOGICAL RECORDS TABLE 28 be read reversed (low variability accom Maxima of Five-Year Averages of Coefficients low mean values) throughout, to achie of Variability Now what is the meaning of this rel Period 2 3 4 5 - 8 dimension magnitude and variability? that when fashion brings a given trait 1787-91 34 40 1792-96 ........ 1797-1801 11 17rl IIII1 311 4II 1 5I 7I 81 9' IT 1802-06 M V 1807-11 .. 12 37 2 1812-16 .......... 4 . 4 1817-21 . .. 1822-26 . 1827-31 11 A 1832-36. 4 1837-41 11. . 11 1842-46 . 1847-51 . 3 1852-56 .......... 1857-61...... 1862-66 . ..........X e2 1867-71 . 1872-76 . o 1877-81. 1882-86. 1887-91 .......... 1892-96...... 1897-1901 .....MV 1902-06...... 1907-11...... 1912-16 ...... 0 1917-21 .......... 5 44 1922-26 ..... . 16 1927-31...... 1932-36 ......32 20 29 Much the same appears from the stars and X 20 double daggers in the ten-year table 26. X I\ As before, high variability tends to be as- X o / \ sociated with extreme, of dimension, but not con- v v sistently so. The reason for the inconsistency 5 is in this case clear, and will be the next XQ 0 point discussed. It occurred to us to plot together the di- mension means and their variability coefficients Fig. 8. Relation of variability (V) on scales calculated to bring out such simi- width (5)o 1787-1936. larity of course as they might or might not possess. Five-year averages were used to plot skirt and waist, ten-year for decolletage. certain magnitude, the style is harmon Figures 8 to 10 show the results. well-knit on that point, and individua It is clear that in four cases out of six, tions, or designs, are in close concor and mainly in a fifth, there is a definite and versely, when this magnitude is depart surprising relation between large dimension and the style is under strain as regards t low variability; conversely, when the dimension ture, and efforts are made simultaneou shrinks, the variability goes up. This is very cede from the magnitude attained and to. conspicuous for both skirt and both decolletage beyond it. In other words, from the diameters (Nos. 2, 5, 4, 8; figs. 8, 10). It underlying pattern of style, there se holds also fairly well for waist length (No. 3; an optimum magnitude or proportion for fig. 9), except before 1821 and after 1921 when ture, when variability is low, and the it rev6rses.23 Waist width (No. 7; fig. 95 must concurred in because it is felt to be - 3 There appears no reascn wvhy this etp 3In No. 3 of fig. 9, V above 10 and M below should not be applicable to the minori 20 or above 32 have been indicated by extra* **** blackness of line, to emphasize that before 1821 in which low variability accompanies 1l and after 1921 the variability reacts to extreme values. That is to say, in most of ou means in opposite manner. the basic style is felt as satisfying, RICHARDSON AND KROEBER: THREE CEINTUIES OF WOMEN'S DRESS FASHIONS 145 able, when the silhouette dimension is and shoulder exposure too ample in evening ut in other traits, when it is small or dresses, to satisfy the ideal of the style. However, we have not only this basic pattern is interpretation we can construct a or ideal style, which is aesthetic with a ting- ideal pattern of Occidental women's ing of the erotic, but also a concept of tempo- or formal dress during the past 150 rary mode or fashion as such, which demands change, It has a long skirt, ample at the bot- and, when it has exhausted the possibilities of erpanse of bare breast and shoulders, material, color, and accessories, goes on to al- and wide as possible, although for me- ter the fundamental proportions, in other words treasons only one diameter can well be the basic aesthetic pattern. With such altera- in at the same time; as slender a waist tion there comes strain, simultaneous pulling ble; and a middle or natural waist-line forward and back; violent jumps in opposite di- between 22 and 30 by our scale; when rections within one or two or three years, and t line gets beyond these limits, and heightened statistical variability. ither the breasts or the hips, the basic The several proportions are successfully at- 4Is violated, resistance and extravagance tacked and distorted by fashion at somewhat dif- oped, and the variability rises. ferent times, and hence the picture is compli- it differently, a confining corset cated. Nevertheless, there emerge periods of a ,omfortable to the wearer, but it is generation or so when fashion is particularly nesthetically satisfying by Europeans active in its attempts to break up or pervert at century and a half, even if it con- the basic pattern. Such are the decades 1785 to aturally, provided it comes at or 1835, and 1910 to the present. Between them, natural waist. Skirts on the other there lies a longer period of essential agree- t be too full or too long, and breast ment and stability and low variability, in which 1111 311 5 61 711 91 11 I 11 131 :1 17 9618061 26 T 6 7 1 3? 9 T If 2 3 .3 _ 3 4 - j27 - e - ' A 123 - i12.1 V~~~~~~M 7 -. - .~~~~~~~~~~~~29 8 ,i.2-~~~~~~~~ !W2 ) ,18- 6d 7 - -~ j - J~I7 \V /\ /J t ~~~~~~~~~D11 V 1 - -~~~~~~~~~2 l i i i I I I I I I I I I I I I I I I I 1 I I I I I I I II 9. eltin f'vaiailty(V)an apl- ig 1. eltin o vribiit () ndamli 146 ANTHROPOLOGICAL RECORDS fashion accepted, or fulfilled, the pattern 2 5 3 7 4 8 while modifying it in superficial detail. 1620- We have too few data to compute variabili- ties before 1787. This is unfortunate because 30- 0 most of the eighteenth century evidently re- 0 sembled the middle and late nineteenth in hold- 40- ing fairly close to what we have determined as 0 0 the basic pattern: the skirt full and rather 50- 0 long, at least not markedly short; the waist, if not narrow, at least accentuated, and in 60- median position; decolletage considerable. If 0 our hypothesis holds, the bulk of the eighteenth 70- century should accordingly.prove to be a period 80- 0 of low variability, on assembly of sufficient 8 data.' ofvaibiite However, we can make the trial assumption 90 that the specific associations of variabilities DATA with crests which we have found to hold since 1 D A Clu 1787 also held before that date, and see how the 10o JO ext results plot out. That is to say, while we have X lat no reliably computable variabilities for most of 2 bi} the seventeenth and eighteenth centuries, we do j So; have fair approximations to the points in time 30- i eas at which the maxima and minima of dimensions o Omen fall; and by plotting the maxima and minima for 40- 7;' the whole three-hundred-odd years, we may hope D cre to discover whether the pre-1787 period shows a 50- ass tendency toward clustering of crests comparable > low to that after 1787. The result of the experi- 60- tio ment is shown in figure 11. Solid circles show te those dimension crests, whether maxima or minima, 70- < (no which since 1787 have been associated with low n dat variabilities and pattern stability; hollow 80- 0 * n circles, crests with the opposite association. 0 -Ho1 The diagram makes it evident that there was 90- * 0 Ho a clustering of crests between 1630 and 1680. a men Seven of the ten crests in this period fall be- 1800 M 4, tween 1645 and 1665. 0 in. 10- ~ ~ ~ ~ ~~.cre~ However, the ten crests are nearly evenly l0 0 ? as divided between those hypothetically associated 0 0 hi0 with high and with low variability, and the two 2 O ti kinds are interdistributed scatteringly. While 30 > te we may accordingly infer that the mid-seven- ea' teenth century was a period of attainment of 40J a style dimension extremes, and rapid alternation c0 va between extremes in at least some features, _ there is nothing to prove that the particular ( post-1787 associations of one of a pair of ex- 60- * * 241t would also be desirable to try to define 70* 0 the basic pattern of dress by inclusion of more features than the six so far dealt with. The 80- treatment of the arms, bust, and hips, in the basic pattern, have not been considered at all. There are important traits here: sleeves; promin- 90- ence and position of the bust; proportion of the hips to shoulders, bust, and base of skirt-- 60O- compare for instance the Grecian bend and bustle fashions with the recent one of hips larger than 10- base of the skirt. But the difficulties are considerable in dealing with these features over 0 longer ranges of times: some disappear and re- 20- 0 appear, others require profile views for full0 0 0; measurement. Nevertheless something could no 30- 0 doubt be ascertained by further analysis. 0 0 RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 147 with heightened variability already held and dress to dress; and shorter periods in which eventeenth century. basic pattern is disrupted or transformed, ex- follow nearly a hundred years, from 1680 tremes of proportion are numerous, and high vari- with but two peaks, in dimensions 5 and ability prevails. on allowance for there being no data This differentiation of periods is positive -1710, there remains a three-quarter- in all respects for the era since 1787. It holds [span with but these two crests. Under as regards stability of basic pattern and infre- ading stability, variability may be quency of extremes for the hundred and eighty to have been low. years preceding. Whether it also holds for the uext forty years, 1778 to 1817, show association of variability with those extremes crests. Four of these are of the type which conflict with enduring stable patterns, ed as of low variability, seven of high. we have not the evidence to prove or disprove; ,the low-variability ones fall mostly but at least there is nothing in the imperfect the first half of the period (mean date pre-1787 picture to argue against such variabil- the high in the latter half (mean date ity association. following eighty-odd years, to 1900, nine crests, seven of them of low- CAUSALITY OF CHANGE lity type. The period seems to consist spans. First, some two decades, 1825- We are now in position better to weigh the \quieting down from the preceding turbu- several possible causes of changes in variability. Then a long calm, only slightly ruffled The primary factor would seem to be adherence low-variability crests (and one high) to or departure from an ideal though unconscious 1860-69 decade, and none at all for pattern for formal clothing of women. The con- ecades after. This Victorian era was sistent conformity of variability to certain mag- ly placid in fashion. nitudes of proportion--mostly a conformity of low 1900 to date, there are ten crests, the variabilities to high magnitudes--leaves little liest of low variability, the last six room for any other conclusion. Or we might say that 1903-17 was a A second possible explanation, that high vari- which variability was increasing but ability is a function of extremes of proportion, v-variability maxima were reached; it falls as such. It is true for a full waist and final phase of the preceding stable period: a narrow or short skirt, untrue for slender waist was already manifest but reaffirmations or full or long skirt. The explanation holds dying pattern were being made. The period only so far as it is subsumed in that of the *, by contrary, shows in every feature basic pattern. d a crest which is in extreme opposition A third possible explanation, that generic or prevailed during the long Victorian calm nonstylistic factors unsettle fashion at certain in each instance accompanied by very high times, is not eliminated, but is pushed into the hity. Table 29 summarizes this. background of further investigation. After all, such a cause would be an ultimate, not an immedi- TABLE 29 ate one. It may well be that unsettled times make for unsettled styles. Revolution, Napole- s of Dimension and Variabilitv by Peniod onic and World wars, struggles over the rights of man, Communism and Fascism, the motor and jazz, od Dimension extremes may contribute to fashion's trying to stretch and ed) Years Low High Total Per disrupt its fundamental stylistic pattern. But - v v Twhile such an influence is easily conjectured, 680 50 (6) (4) 10 2.0 it is difficult to prove. In any event, there 777 ... 97 (1) (1) 2 0.2 seems no clear reason for the specific fashion 817 ... 40 4 7 11* 2.7 extremes which such a set of causes might be 902 ... 105 7 2 9 0.9 thought to produce. Social and political un- 934 ... 32 4 6 10 3.1 settlement as such might produce stylistic un- variability type of three crests before settlement and variability as such; but there is e variability type of three crests before nothing to show that it would per se produce thick waists, ultra-high or low ones, short and tight skirts. If there is a connection here, it is clear from this table as well as figure seems that it must be through alteration of the tthere occur in European women's dress basic semi-unconscious pattern, through an urge Lting longer periods in which a basic pat- to unsettle or disrupt this; and that when in- Ofstyle is rather stably adhered to, rela- creased fashion variability occurs, it is as a few extremes of proportion or dimension direct function of pattern stress, and only in- liht, and those all in a direction accom- directly, and less certainly, of sociopolitical by only low variability from year to year instability. In short, generic historic causes 148 ANTIROPOLOGICAL RECORDS tending toward social and cultural instability characteristic of the agitated periods is may produce instability in dress styles also; so much extremes of dimension or proporti but their effect on style is expressed in stress extremes of high variability; and these in upon the existent long-range basic pattern of correlate with certain minima and maxima 0o dress, and the changes effected have meaning portion, but not with their opposites. only in terms of the pattern. nificaant fact remains that high variabili Concretely, it would be absurd to say that not associated with any dimensional crest, the Napoleonic wars, or the complex set of always26 with only one of a pair of oppos historic forces underlying them, specifically tremes. This throws us back on the basic produced high-waisted dresses, and the World War tern as something that must be recognized. low-waisted ones. They both probably did pro- Now, one can indeed accept this basic p duce an unsettlement of style, which, however, tern, but accept it as something intrinsic resulted in extremity of high and low waisted- tending to remain more or less static over ness respectively.25 long period, or the whole of a civilizatio Herewith arises another question: whether the then attribute the more marked variations crests and troughs of waves of fashion, its peri- it to broader historic disturbing causes, odicities discussed in section IV, are perhaps than to anything stylistically inherent also to be sought not in anything inherent in tending from within toward swings away fr fashion, but rather in more general historic back toward the pattern. On this view th causes. In favor of such a view is the heavier tury-long cycle which we have found to ho clustering of trait extremes in Revolutionary- most of our fashion traits would not be a Napoleonic, World-War, and immediately subse- erty of style per se, but a by-product of quent decades. But again there are crests also fact that Europe happened to be generical in the intervening period. What is specifically turbed in the decades around 1800 and 192? VIII. CONCLUSIONS Our first finding is that the basic dimen- made from them, but they are not tampered sions of modern European feminine dress alter- except again unconsciously. nate with fair regularity between maxima and This in turn seems to imply that the r8 minima which in most cases average about fifty particular individuals in molding basic years apart, so that the full-wave length of style is slight. The influence of creati their periodicity is around a century. important individuals is probably largelyi By comparison, annual changes, and even those on the accessories of transient mode. Ho of moderately long periode of moderate length, it is there, has never been objectively e generally are markedly less in degree or ampli- and would be difficult to investigate. Hi tude. This conclusion applies to the major pro- ans of fashion may be partly right or mai portions of the total silhouette. Superstruc- fictitious in the influence they assign to tural features have not been examined quanti- Antoinette, Recamier, Eugenie, and the var tatively, but appear to develop and pass away Princes of Wales. The reverse is much mor completely in briefer cycles. The present study likely, that individuals conform to the st is concerned with the variations in persistent which they find in existence, operate in features. ways within its configuration, and at tim There appear accordingly to be two components coincidence receive false credit for "caus in dress fashions. One is mode in the proper one or more of its features. sense: that factor which makes this year's 25The Empire mode was consciously Greek! clothes different from last year's or from those Classic. It professed to take over from of five years ago. The other is a much more quity a full and rather high waist and a f. stable and slowly changing factor, which each instead of flaring skirt. It obviously di year's mode takes for granted and builds upon. take over from antiquity its own short ski It cannot be pretended that these two factors wide decolletage, nor its ultra-high waist a tight, undraped skirt, nor short puff-sl are definably distinguishable throughout. Be- In brief, Empire dress style fell in with havioristically, however, they can mostly be catchwords of its day, and in consonance w separated by the length and regularity of the the social currents and political currents changes due to the more underlying component. its time, which aimed toward the Classic, Itnis edent tha the basic far ,pofent cepted just as much of Classic dress style,: . t S eveenttnattte aslctea-Jure otsuited its own trends, and for the rest fo style as distinct from more rapidly fluctuating these trends while calmly ignoring or viol mode, being taken for cJranted at any given no- all the remaining features of its supposed meat, are largaely unconscious in the sense that 2Excepting dimension 3, waist length, w they are felt .as axiomatic and derivations are variability is associated with medium nagni RICHARDSON AND KROEBER: THREE CElNTURIES OF WOMEN'S DRESS FASHIONS 149 long swings of proportion which we have therefore a relation between extremes and vari- Lned seem comparable to what economists ability. oular trends, which also carry oscilla- However, this relation is one-sided. For ir lesser cyclic movements on their sur- four of the six proportions examined, variability No one attributes either these larger rises as the proportion or diameter shrinks, be- ia trends or the fluctuations to individual comes low as this reaches ampleness. For a fifth ive. It is of course conceivable that proportion, waist width, the relation is the op- c determinants are social in their nature posite. For the sixth, variability becomes acute listic ones individual. In fact this is when the measure is either very high or very low. saumed. However, such an assumption is High variability thus is more completely in the sense of being critically untested. limited to certain periods than are extremes of ather more likely that what holds in one the proportions or diameters themselves. Those of human culture holds also in another. of the diameter extremes which are accompanied rate, the burden of proof must rest on by low variability fall in some cases into the trary view. And this burden has cer- long stable interval. doubled since we have shown that dress- The best explanation that we are able to sug- ges behave historically somewhat like gest for these phenomena is that of a basic pat- e ones; in the stateliness of their march, tern of women's dress style, toward which European , for instance, and in their superim- culture of recent centuries has been tending as :ycles or oscillations. an ideal. This pattern comprises amplitude in evidence to date shows that when a pro- most dimensions, scantness or medium value in has swung one way to its extreme and others. As these proportions are achieved, there lfway the other, it may oscillate for a are equilibrium, relative stability, and low vari- or two part way back to the first ex- ability. The pattern may be said to be saturated. but normally it resumes its swing toward At other times, most or all of the proportions osite. But this is a behavioristic find- are at the opposite extreme, which may be con- d a priori may just as well be due to strued as one of strain, and variability rises 1 as to personal causes. So far as in- high. This basic or ideal pattern, for Europe s are concerned, the total situation of the last two or three centuries, requires a overwhelmingly to indicate that their ac- skirt that is both full and long, a waist that are determined by the style far more than is abnormally constricted but in nearly proper a determine it. anatomical position, and decolletage that is am- generic significance can be claimed for ple both vertically and horizontally. ue of a century found for the average The periods of computed high variability and city or wave length of dress proportions. therefore of "strain" or perversion of pattern only a mean, though it is rather closely coincide fairly closely with the Revolutionary- to in three of our six features. Obvi- Napoleonic and World War-post-War eras. Generic other features, or styles other than cultural or historic influences can therefore European ones, may possess quite differ- probably be assumed to affect dress-style changes. iodicities. In fact, there is no reason Sociocultural stress and unsettlement seem to yle in general, or even dress style, produce fashion strain and instability. However, necessarily swing rhythmically back and they exert their influence upon an existing sty- Our findings apply only to the material listic pattern, which they dislocate or invert. d. Without reference to this pattern, their effect initely significant is the fact that there would not be understood. iods of high and of low variability of While we have no reliable variability meas- These come out much alike whether it is ures before 1787, it is clear that in the dec- er of variations of yearly averages from ades surrounding 1650-60 there was an accumula- e-year moving mean, or of variations of tion of proportion extremes similar to those of dual dresses from the year's mean. Within 1787-1835 and 1910-36. The mid-seventeenth cen- t century and a half, 1787-1835 (espe- tury may thus have been a third period of pat- before 1820) and 1910-36 are periods of tern strain, rapid change, and variability. riability. The intervening seventy-five The explanation propounded is not that revo- years show low variability. The avail- lution, war, and sociocultural unsettlement in asures scarcely allow of variability com- themselves produce scant skirts and thick and ons for most of the eighteenth century, high or low waists, but that they disrupt the general pattern apparently underwent no established dress style and tend to its over- rked alterations in that century until throw or inversion. The directions taken in 775 ~~~~~~~~~this process depend on the style pattern: they two high-variability periods also con- are subversive or centrifugal to it. By con- ore crests or extremes of proportion than trary, in "normal" periods dress is relatively intrvening seventy-five years or than the stable in basic proportions and features: its Fie bulk of the eighteenth century. There is variations tend to be slight and transient-- 150 ANTHROPOLOGICAL RECORDS fluctuations of mode rather than changes of is no need to fall back on assumptions ofA style. In another civilization, with a differ- unknown factor inherent in dress itself ent basic pattern of dress style, generic socio- ing for rhythmic change. cultural unsettlement might also produce un- We have deliberately avoided explanati settlement of dress style but with quite dif- our phenomena in terms of psychological fa ferent specific expressions--slender waists and such as imitation, emulation, or competiti flaring skirts, for instance, or the introduc- which are a stock explanation: the leaders tion or abolition of decolletage. to surpass the mass, so they keep going O It is conceivable that the method pursued in step farther, until a physical limit is r this study may be of utility as a generic measure when they turn about and head the processi of sociocultural unsettlement. Also, it pro- back. We do not deny that such psychologi vides an objective description of one of the motivations may be operative. We do belie basic patterns characteristic of a given civili- that as explanations they are conjectural,. zation for several centuries, and may serve as a scientifically useless, because, to date a precedent for the more exact definition of other least, they depend on factors which are stylistic patterns in the same or other civili- urable and undefinable. On the contrary, zations. think we have shown that through behaviori It also seems possible that the correlation and inductive procedures operating wholly with general conditions explains the near-regu- the sociocultural level, functional correl larity in the periodicities of dress. If these tions can be established for such supposed largely express pattern disturbances due to refractory cultural manifestations as styl disturbances more general in the culture, there fashion changes. APPENDIX BY A. L. KROEBER ort and Hartman27 have analyzed the social control. It is also a "genetic" approach of my paper of 1919, along with Chapin's in the sense that it deals with unique histori- Cultural Change. Some of their criti- cal phenomena not subject to experimental veri- ill presumably be raised, by them or fication. How far the approach is "stylistic" against the present study. It seems in the Allport-Hartman sense, other than that re worth while to consider their argu- the material dealt with concerns style, I cannot say, because I do not understand their definition t of all, it is significant that Allport of this approach. It seems rather a subdivision t{an call their analysis "The Prediction of the "genetic," in so far as styles are his- ural Change."28 This was certainly not torical phenomena. In fact, is not the primary purpose. I find only two predictions concern of history and cultural anthropology, so paper; and neither of these is material.29 far as they transcend the writing of individual affirmations of emphasis of my convic- biographies, precisely a dealing with "styles" t there is "order" or "regularity" in of human behavior? The one thing that my paper >ohages. If it will clarify the issue, was free from, as I see it, was telic or practi- be glad to withdraw both predictions cal approach. I may be wrong here: one is a poor ing been unnecessary. judge of his own motivations: but it is also pos- the emphasis of the analysis is so strong sible that Allport and Hartman have started out iction, is hard to see, especially as with preconceptions which have caused them to 's Cultural Change also is concerned pri- misunderstand my objectives, underlying and pat- with change and not with prophecy. The ent. At any rate, so long as theirs is the only e explanation is that the analysts, in formal criticism which my article has evoked, I ,with most sociologists and economists, should like to go on record. selves interested in the future, in the Where I have presumably given provocation is bal consequences or applications of study, in the loose use of deterministic terms such as t are so committed to what they call the law (usually in quotation marks), principle, n approach, that they assume it and the cause, order, and regularity. If it will clear o approach to be the only ones possible the atmosphere toward understanding, I will field of social data. cheerfully retract any of these that are confus- eems to me that my approach in the former ing. There is no use quarreling over metaphors. was a "natural science" one in the sense That there is a certain "order" or "regularity" was empirical, inductive, objective, in the phenomena, enough to prevent their being e of any motivation of applicability or construed as due to the caprices of individual human wills, I continue to believe, and think oyd H. Allport and Dale A. Hartman, The Dr. Richardson and I have overwhelmingly proved tion of Cultural Change: A Problem illus- in the present paper. "Determinism" in this in studies by F. Stuart Chapin and A. L. sense I adhere to: that the actions of individ- pp. 307-350 of Stuart A. Rice, editor, ual persons are determined much more by styles s in Social Science: A Case Book, 1931. and other sociocultural influences than they de- t this is not a mere matter of an editor's termine them. As to "laws," it seems perfectly is shown by the analysts' statement on cermfro the to whee perectly 16: "The central question of Professor clear from the context even where there is no 's research concerns the possibility of explicit qualification, that I was not claiming tion from this linear stylistic approach"; to have found laws comparable to those of phys- their devoting nine pages to an analysis ics. However, let us consider all statements natural-science, genetic, telic, and on this score as withdrawn. io approaches with reference to their The argument about the sense in which winter tive predictive value. Termine ng se nse inrwhistic 249-250: "By 1912 the tide has once "determines" spring seems another verbalistic ed--no doubt to continue now for an- one. We happen to know the astronomical caus- two or three score years unless the peri- ality of our seasons. If we did not, we would y of the rhythm is accelerated. . . ." still know empirically the regular order of the recast has been proved premature by the swing of the seasons; and if any medicine-man, 1911-12 was indeed a low point for a wdth, and the amplitude increased until after spring had followed winter, asserted or but the narrowing resumed until a still was credited with bringing on summer by magic Point was reached in 1926, which pre- subservient to his personal will, a very limited rymarks the real trough of~ the wave, natural science experience would protect us 0t phranin 919 sThits iis correct since against such superstition, even if it led us to For anyone setting out to be a prophet, talk a bit loosely about a "law of the seasons." rification out of two would be a sorry In the field of cultural happenings, it is patent . ~~~~~~~~~~that we do not know at all what the real causality i ~~~~~~~~~~[151] 152 ANTHROPOLOGICAL RECORDS is. '!e also know very little about such order lysts is through such factors as changes i 1-1d regularity as there may be. I was trying economic and social position of women, 8sO to make a contribution, at one concrete point mode of dancing, cost of materials, and su where the data looked promising, toward ascer- These are cultural factors, such as I beli taining what order could be found in cultural should be correlated with stylistic change phenomena--as did Chapin in his Cultural Change. fore the last ditch of psychological expi The alternative to rejecting such endeavors is tion is retreated to. However, they are s to hold that history is a series of accidents. cific, immediate, and temporary factors wv This in turn, since most human minds will not primary effectiveness seems to be largely remain content with utter negativism, usually cluded by the long range of the swings. results in a reversion to the popular assump- there is a suspicion of preference for the tion that cultural events are produced by the ordered interpretation: style may not be a free volition of personalities. This assump- cious, but it is accidental; the aoncept of tion, again, irn twentieth-century scholars, superindividual pattern is avoided and resi seems nearly as naive as the belief that di- Now why this resistance? why all this rected macic can produce summer. reling with my language, the imputation of If it is mere philosophy or mysticism to be- ticism to a quantitative investigation such lieve that culture determines the actions of economists conduct by the thousand, the lo personalities, at any rate determines them far and partly irrelevant wind-up about scienti ,ore h-han their uncontrolled volitions determine and telic approaches, the near-pedantic pie culture, then I am a mystic. However, the long, on the statistical reliability of my data i persistent swin-gs of style seem an empirical the last paragraph?30 fact which does directly support this belief. I susDect that the resistance goes back How do Allport and Hartman meet the very real the common and deeply implanted assumption and fundamental issue that there is a clear-cut our wills are free. As this assumption has finding here which rests on evidence? to yield ground elsewhere, it has taken ref They fall back, tentatively, on a psycho- in the collective, social, and historical s logical explanation: emulation, which leads each Since the chemists, physiologists, and psyc individual to try to outdo the rest until a gists have unlimbered their artillery, the physical or other limit is reached, when they sonal freedom of the will is thankless terr all race back competitively. Interestingly to maintain. Culture they have not yet att enough, this very explanation of emulation was so that becomes a refuge. Whatever the de adduced nineteen hundred years ago by Velleius to which we have ceased to assert being fre Paterculus when he wrestled with the problem of agents as individuals, in the social realm why most high cultural activities seem to come can still claim to shape our destinies. Th- in cyclic bursts. Now, that a view is old does theologian is piping pretty small, but the not prove it unsound; but if twentieth-century cial reformer very loud. We are renouncing sociologists can do no better than to speculate kingdom of heaven, but going to establish a like a soldier-historian contemporary of Augus- near-millennium on earth. Our personal wil tus--well, they also have not traveled much may be determined, but by collectivizing th farther than the mystics. Of course there is we can still have social freedom. psychology involved. Every cultural event in- Of course this is not conscious motivati volves psychological happenings. What my an- But I see no other motivation for the resis alysts fail to realize is that they do not, just to yielding an inch to any form or degree o as Paterculus did not, connect their psychologi- cultural determinism. cal explanation with the cultural phenomena by The amount of determinism shown is reall any evidence. They have made a guess, an un- very little, in the original essay: no raore verifiable conjecture. This is fine for after- than that there are stylistic trends of an dinner conversation before the fire, where too plitude, effectiveness, and duration indica much evidence destroys social affability. I that they are governed by factors which are may be mystic or telic, but I have at least tried to relate observed phenomena with observed 30 phenomena.~~~~~~~~~~ I may ad-a mycnito hc The tenor of the original article makes, phenomena. I may add--as my conviction which I plain that there is no insistence on the re cannot prove--that my guess is the same: I also ability of any one figure or group of figur believe that emulation or imitation is involved but only on the duration of swings and peri in style changes. But I have deliberately left tence of trends. The absence of sigmas and it out of my interpretation. Paterculus sav., a probable errors is cited as "a vital defect. it out Of my interpretation. Paterculus saw: a As a matter of fact the changes found are so very real and still unsolved problem in a large great that if we had data on only a single n and important group of pheno-;iena. But if we for each year from 1844 to 1919, instead of cannot attack the problem any better than by his to ten, the fact of notable swing in several nonevidential method, I for one will call it dimensions would be convincing to common se . q ~~~~~~~~~~~and, I suppose, provable as statistically si quits and play another game. nificant if one were minded to show his tech The other tentative explanation of the ana- nical virtuosity. RICHARDSON AND KROEBER: THREE CENTURIES OF WOMEN'S DRESS FASHIONS 153 but which must be superindividual: a ran- phenomena. We also do not deny that sports or ies of free wills could not pull together the invention of rayon may have an effect on direction so long and decisively. Apart dress styles. We have left them out of the i presentation and analysis of the data, reckoning, for the time being, because their L9 paper is essentially a statement of influence is presumably special and limited, sis--no doubt reiteratively and at times and therefore secondary to the major swings of e looseness; but it is difficult to be dimension and variability. We are aware that a and exact when one is compelled to swim "general historic unsettlement" involves psy- the intellectual current. chological attitudes as much as emulation in- he present monograph I believe that Dr. volves them; but they are attitudes which are on and I have gone a little farther. We at least partially measurable and definable in rengthened the case for authenticity of terms of wars, changes of boundary or form of sg8 by considerably enlarging the range of government, abolition of old institutions, new tion. We have also examined the vari- codes, intensity of class struggles, aesthetic rof the phenomena and found this to cor- innovations, and the like; and historians are definitely both with certain periods of agreed in recognizing these phenomena as having and with certain statuses of the swings. occurred. The stone we have tried to lay on is double correlation we have inferred as the wall of the understanding of the history of , for the last two centuries of Europe, human civilization may be quite small; but we imo between periods of general socio- feel that it is at least tangible and weighable al and cultural unsettlement and periods as evidence. mity of fashion, through the medium of I realize that any interpretation which di- ssible and approximately definable pat- minishes the range of free personality and en- dress style. We have not fallen back larges the effectiveness of superpersonal cul- tion as an explanation of specific tural influences is likely to be unpalatable. It not because we deny its influence, but will irritate many and it will elicit rejections. we do not know any specific way of cor- But I am compelled to adhere to it--no doubt by emulation with particular historical the strand of culture of which I am part.