America measured 14,698 women to fix dress sizes — then chose labels over measurements

In the late 1930s American women's clothing carried no standard at all. The government study that created the system opens flatly: there were no standards for garment sizes, sizes had grown up by trial and error from measurements of a few women, garments labeled the same size varied greatly between manufacturers, and alterations could add as much as 25 percent to the cost of a dress. Mail-order houses, which could not offer a fitting room, absorbed a large volume of returns — which is what pushed the Mail Order Association of America toward funding a fix.

With a Works Projects Administration grant, statisticians Ruth O'Brien and William C. Shelton organized state units that measured 14,698 women across the country, taking 59 measurements each, and published the results in 1941 as USDA Miscellaneous Publication 454. The data destroyed the industry's founding assumption: girth and height were nearly independent, so a bust-based size could not predict a fit. A 46-inch-bust woman is no taller, on average, than a 32-inch-bust woman. Fit required a two-dimensional classification of body types — and a grid that big threatened to be unusable.

The decisive move was the label. O'Brien and Shelton wrote that the nomenclature 'would probably have to be on some arbitrary basis, as is the present system of shoe sizes, rather than representing the numerical values of some anthropometrical measurement.' When the National Bureau of Standards reanalyzed their data, it issued the 1958 commercial standard: bust-based sizes 8 to 38 with height and girth modifiers. Sears and Montgomery Ward adopted the government sizes, and numbered sizes became the shared language of American clothing — built on labels that no longer pretended to be inches.

Arbitrary labels solved adoption and planted the decay. Because nothing tied a size number to a body, manufacturers could quietly re-map the labels — vanity sizing — and by 2008 an ASTM size 8 matched a 1958 size 14 or 16. The government standard was downgraded to voluntary in 1970 and withdrawn in 1983, leaving the 1941 taxonomy — misses, junior, petite, tall — still standing under numbers that mean whatever the brand wants.

Make the size label arbitrary, not a measurement

The survey replaced folklore with data: 59 measurements on 14,698 women let the industry classify real body types instead of grading everything from one proportional mannequin.

Arbitrary labels lowered adoption barriers: no manufacturer had to publish measurements, no shopper had to state her weight — the surveyors predicted shoppers would refuse a weight-based label.

A shared grid was worth more to catalogs than a perfect one: any even-numbered size that meant the same thing at Sears and Montgomery Ward cut returns on both sides.

The two-axis structure (number plus height and girth modifiers) encoded the data's actual finding — bust and height vary independently — instead of pretending one tape measure could do it all.

Abstraction can make a standard adoptable: round labels hide exact data from vanity and inertia. But arbitrary labels cut the cord to reality — nothing stops the mapping from drifting, and it did.

The 1958 commercial standard never bound most manufacturers; it became a voluntary product standard in 1970 and was withdrawn in 1983. ASTM has kept the 1958 nomenclature since 1995 while the numbers inflate: a 1958 size 8 fit a 31-inch bust and 23.5-inch waist; by ASTM's 2008 tables it fit a body five to six inches larger — a 1958 size 14. The 1941 data also had a flaw: measurements of women of color were taken but discarded. SizeUSA scans later found five to seven body shapes; the hourglass fits about 8 percent. The labels survive; the numbers mean whatever each brand chooses.