REPORT ON BIRD CENSUSES, 1916-1920. 3 



the selection of the area to be covered, the time at which to make the 

 count, and the information to be included in the report. As a 

 knowledge of methods followed is essential to an understanding of how 

 the conclusions here presented were reached, it will be well to outline 

 what are considered the best conditions and procedure in taking a 

 bird census. In the following discussion the experience gained from 

 the actual work is embodied with the early plans and instructions. 



METHODS OF TAKING BIRD CENSUSES. 



It is obviously impossible to enumerate all the birds in the country 

 or even in a small section, but the birds nesting on a tract of not 

 over 80 acres can be counted by one observer with considerable 

 exactness. The combined results of many such counts will furni>h 

 a key to conditions in the country as a whole. The sample tracts 

 should represent as nearly as possible the average conditions for 

 their respective localities in regard to the proportions of woods, 

 orchards, crops, pasture, etc. It is, of course, impossible to find one 

 tract that contains all of these elements in exactly the right pro- 

 portions, but the greater the number of tracts surveyed, the less will 

 be the chance of error in the resulting averages. 



The plots of farm land chosen for the censuses should contain on 

 the average from 40 to 80 acres. A tract of less than 40 acres seldom 

 shows sufficient diversification to be representative; and an 80-acre 

 tract is as large as one person can accurately cover under most cir- 

 cumstances. When the enumerator lives on or near the land sur- 

 veyed and has the time to devote to the work, larger areas can be 

 and have been very satisfactorily worked. 



Sometimes because of local conditions it is desirable to have a 

 census over a large area. In such cases, especially if the tract shows 

 considerable diversity, the work is facilitated and the interest and 

 value increased by dividing it into two or more sections and treating 

 each as a separate count, thus indicating how the birds are distributed. 

 Contiguity of the tracts surveyed adds materially to the interest and 

 value of each, since not only is a larger area thereby represented, 

 but something of local distribution is shown. Two series of counts 

 from Kansas, elsewhere quoted somewhat at length, are excellent 

 examples of this (pp. 11-12). 



It has been deemed important that, so far as possible, areas be 

 selected on which conditions are not likely to change for several 

 years. Under such circumstances it will then be evident that 

 fluctuations in the bird life on these areas are not due to artificially 

 changed environment. Nevertheless, counts of all kinds are welcomed, 

 for it is impossible to have too many; moreover, the response of birds 

 to changes in environment is an interesting study in itself. 



It is very important that counts be made on the same areas year 

 after year, in order that the records may give some basis for com- 

 parisons. Only by the accumulation of large series of records 

 dealing with the same land can definite conclusions be drawn as 

 to average conditions or as to changes taking place — whether birds 

 generally or individual species are increasing or decreasing. A 

 study of the conditions that existed in the eastern United States in 

 1918 (see pp. 28-31) has shown the value of such series of records as 

 a means of studying fluctuations in bird life. 



