Labor force data must be evaluated in the light of historical evidence about changes in census procedures and in the social context of enumerations. Net growth of the female labor force from 1890 to 1950 was much less than census data indicate. Reported growth was augmented by broadening of census definitions, improvement of procedures, wider awareness of women's employment, increasing willingness to report women's work, and the shift from self-employment and homework to wage and salary employment away from home. Internal checks and comparisons with other enumerations suggest serious undercount of women workers by earlier censuses. The best approach to the trend data is to abandon the uncertain early totals and concentrate on sifting out the more reliable details.