1 96 rim i ips, v | \p\Vi 



b\ a single observer, as much too variable a quantity to bo oi 

 practical use. By bis methods of estimating the dates from year 



to year he obtains an extraordinarily slight arrival variation. 



Thus it is probable that ordinary dates of arrival as given by 

 single observers increase rather than diminish our knowledge of the 

 actual potential accuracy of the species. 



In 'Cassinia 1 for 1911 arrivals in the Philadelphia region are 

 tabulated for the years 1906 191 1. Below are given in a table four 

 species which are remarkable from OUT point of view. The figures 

 are the actual departure in days from a ten year average for the 

 species. They are based on Mr. Stone's 'bulk arrival. 1 



In glancing down the 'Cassinia 1 table one sees immediately that 

 groups of species arriving at nearly the same time are often very 



similarly affected by the season, and will be either late or early 

 rding to the year. Within the same year, however, the early 

 or March migrants may be affected in an opposite direction from 

 the late migrants; thus early species may be late and late species 

 earl\ . for the same season. This is merely more evidence to show 

 that at least some of the observed arrival error is meteorological 

 rather than instinctive; ami external instead o\ internal. 



in the same paper Mr. Stone notes that in computing the ten 

 year averages it is interesting to see how the average of 'bulk 

 arrivals' based on the method given above, coincides with the 

 average of first arri> als at stations where there have been a number 

 of accurate observers. On page -17 is a table which shows the 10 

 observation stations near Philadelphia, with dates oi arrival for 

 9 common birds. The * bulk arrivals' here are either the same or 

 onl\ one day later than the average first arrivals. 



Taking a single speeies. the Brown Thrasher, a bird easily seen 



and almost impossible to identify wrongly, the ten year arrival 



re given in detail. For this period there are 10 stations 



and 22 observers at work. Each of the stations has its own average 



and the greatest error is -f-_ davs at George School, and 3 days 



