294 U. S. NATIONAL MTJSETJM BULLETIN 217 



before and half after that date. Accordingly about May 20 is the 

 mean and median date for recorded first arrivals of all species. The 

 arrivals about this date are not arranged in a normal distribution 

 curve, but form one negatively skewed. For early season progress 

 the curve is long and concave, and after May 20 it is short and convex. 

 Since the peak of arrivals precedes the maximum of summer heat by 

 about two months the relations to maximum heat is not explanatory 

 for migration. 



Influence of Seasonal Phenomena 



The accelerating progress of environmental conditions in spring 

 and a factor resulting from contingencies of observation may both be 

 concerned in the skewed distribution of arrival dates. No bird can 

 be recorded before it arrives, and even with intensive search some de- 

 layed observations inevitably result, tending to skew the curve. The 

 separate plotting of the records for the years 1949-1953 in figure 12 

 shows the component years of the record. The median dates of 

 distribution of arrivals in those years occurred on May 20 in 1948, 

 1949, 1950, 1951 ; on May 24 in 1952 ; and on May 18 in 1953. Observa- 

 tions of arrival were deliberately relaxed in 1952 for the interval 

 from May 21 until May 26 in order to give attention to other studies. 

 On May 29 they were resumed intensively, and it may be suspected 

 that the large block of records in the last of the month resulted from 

 delayed observation. The low temperature and late snow melting 

 in May of that year will be discussed later. It was also the year 

 between 1949 and 1953 when fewest observations were made, and the 

 abridged observation makes it doubtful whether the resulting records 

 well represents the dates of first arrival. With these reservations 

 about 1952, the close correspondence of the median of dates of recorded 

 first arrivals in each year suggests that the composite curve presents 

 a summation of substantially similar annual migration records. 



It might be suspected that the records for those species most 

 consistently observed would contain fewer observational errors than 

 for birds which could be less regularly observed. In order to examine 

 the data by this criterion I have separated in figure 11 the first re- 

 corded arrival dates of species for which 3 or more annual records 

 were obtained. The median date for this group falls on May 19. 

 The small advance is consistent with the view that the most evident 

 birds are reported a little more promptly than those which are less 

 conspicuous. Altogether these views of composite and selected records 

 suggest that the homogeneity of the data results from regularity in 

 bird behavior and has been little modified by erroneous observation. 

 Because of the skewed characteristic of the curves it is likely that 



