144 BUREAU OF AMERICAN ETHNOLOGY [Bull. 197 



were employed suggests that the age distributions from which these 

 low medians were calculated are seriously biased. In the case of the 

 1960 census, we can only surmise that an increased proportion of the 

 adult population was no longer residing in the area. 



The 1936-38 data were collected by fieldworkers in connection with 

 the Human Dependency Survey, conducted by the Soil Conservation 

 Service. These fieldworkers received stringent instructions in regard 

 to the recording of age data for the Navaho. The common estimation 

 procedures whereby the enumerator "assists" the respondent in pro- 

 viding a plausible age were reduced to a minimum, with tlie result that 

 the ages of only about 80 percent of the population covered in the 

 survey were reported.'*^ It is plausible to infer that persons of un- 

 known age would be somewhat older, on the average, than those of 

 known age, so that the medians derived from the population of known 

 age would have a downward bias. 



A somewhat different bias, but in the same direction, is apparent in 

 the data obtained from the population of the Fort Defiance subagency 

 in 1956-57. These data were collected in connection with the first of 

 a series of school censuses carried out in every subagency of the 

 Navajo Reservation. As their designation implies, these censuses were 

 designed to obtain a complete register of children of school age or 

 nearing school age. It is plausible to conclude that in focusing upon 

 the school-age population, these researchers (most of them teachers 

 and other school personnel) would be more likely to overlook older 

 persons, especially if these persons were living alone or without school- 

 age children. 



The highest medians obtained from this group of Navaho age dis- 

 tributions are those calculated from sample data obtained by me from 

 the census office rolls at Window Rock, Ariz., m 1957. These medians, 

 averagmg close to 26 years, are clearly out of line with all of the other 

 medians calculated for the Navaho, none of which is over 20 years. 

 These high medians for the enrolled population in 1957 are a striking 

 indication of the extent to which vital events occurring among the 

 Navaho m the period from 1939 to 1957 were either not recorded on 

 the roll, or were greatly delayed in the recordmg.'-** 



If we assume that the actual enrolled population in 1957 was 

 approximately 10 times as large as the sample figure given in table 28, 

 the total enrolled population would number about 77,000 at this time. 

 An estmiate of the U.S. Public Health Service for July 1, 1957, places 



*^ In table 28, the number of persons of known age, expressed as a percentage of the total 

 population, is in the upper 90's for all of the age distributions shown, except that of the 

 Human Dependency Survey, which was only 80 percent. This lower figure is more in line 

 with what would be expected in view of the relatively high proportion of illiterates among 

 the Navaho. 



^ The population figugres shown in tables 28 and 31 for the enrolled populations In 1939 

 and 1957 are from a sample of approximately 10 percent selected by me from the 

 census ofllce rolls at Window Rock, Ariz., during the summer of 1957. The sample selection 

 procedure is described in the Appendix. 



