154 FRANK LORIMER 



computed on an age-specific basis, has probably dropped to zero during 

 the present calender year, 1932. In other words, the population increases 

 which will still be recorded for several decades must be attributed to pecu- 

 liar age-distribution conditions which are the hang-over of vital conditions 

 that are no longer operative, and to whatever immigration there may be. 



There are three principal types of data which may be used as indices of 

 natality. Birth rates, especially age-specific maternity rates, are probably 

 the most reliable in making a comparison between racial and nativity groups. 

 Birth rates, however, are quite unreliable in making comparisons between 

 different localities, especially between rural and urban areas, because many 

 rural women bear their children in city hospitals. 



Ratios of children under 5 years of age per thousand women of childbear- 

 ing ages, sometimes taken as 20-44 and sometimes as 15-44, give the best 

 basis for comparisons between localities, and may be used for comparisons 

 between whites and Negroes and between different urban districts. Both 

 birth rates and ratios of children to women are subject to inaccuracies due 

 to under-registration of births and under-enumeration of young children. 

 In some sections and with some racial groups such inaccuracies may amount 

 to as much as 25 per cent and in some cases even more. Ratios of children 

 to women give an index of natality which has, in a sense, been automatically 

 adjusted by the elimination of much of the infant mortality. 



The third chief index of natality is size of family, or record of numbers of 

 children ever born to the mothers of a given group. Data of this type may 

 be the best or the worst. Confusion frequently exists between numbers of 

 children ever born and numbers of children counted as alive at any time; 

 and unless the data are standardized for age distribution of mothers or 

 limited to records of women at least 40 years of age, or unless age disturb- 

 ances have been in some other way eliminated, data on size of family is quite 

 worthless for our purposes. Incidentally, it may be hoped that the Census 

 will insert again the question, "Number of children ever born," which was 

 included in 1910 but dropped in 1920 and 1930. Data secured by this ques- 

 tion constitutes the basis of the elaborate studies of fertility which have been 

 made in England and some other countries, and recently in this country by 

 the Milbank Memorial Fund. In recent English and American studies of 

 this sort it has been necessary to use 1910 (England, 1911) census data. 



In the analysis of survival rates life-table data are fundamental. Life 

 tables give the proportion of survivors to different age periods, the expecta- 

 tion of life at birth, and the numbers of persons of different ages who would 

 be expected to be alive if birth rates remain constant and the effects of mi- 

 gration are neglected. All of these values are derived from age-specific 



