386 YEARBOOK OF THE DEPARTMENT OF AGRICULTURE. 



according to an observer on the eastern shore 5,000 dead birds to the 

 mile were strewn along the sands only a part of those that perished. 

 The frequency of these disasters proves that birds can not foretell 

 the weather. No bird starts on a migratory flight during a rain or 

 in a dense fog or against a chilling blast, and yet thousands of birds 

 each year are found near lighthouses and along the shores of large 

 bodies of water under just such weather conditions, showing that 

 after starting they met or were overtaken by the storm. The early 

 settlers in the Mississippi Valley noticed so often that an exception- 

 ally heavy flight of ducks and geese moving straight south at a high 

 altitude was soon followed by a severe storm that they came to have 

 great faith in the birds as weather prophets and believed that they 

 could actually foretell an approaching tempest. It is more probable 

 that the birds began to migrate at the first signs of the storm and 

 outstripped it in their southward flight. 



BEGINNINGS OF MIGRATION. 



It may be safely stated that the weather in the winter home has 

 nothing to do with starting birds on the spring migration, except 

 in the case of a few, like ducks and geese, that press northward as 

 fast as open water appears. There is no appreciable change in tem- 

 perature to warn the hundred or more species of our birds which 

 visit South America in winter that it is time to migrate. It must 

 l.)i a force from within that makes them spread their wings for the 

 long flight. The most important duty of the individual bird is the 

 perpetuation of the species, and the impulse which annually starts 

 the bird north toward its breeding grounds is physiological. 



UNIFORMITY OF ARRIVAL. 



Were the surface of the earth level and the climate absolutely uni- 

 form, birds Avould arrive at a given place on approximately the same 

 day each year, but the records for a series of years at any given 

 locality show considerable variation in the dates of arrival. Part 

 of this variation is undoubtedly due to errors of observation, for 

 series of notes on the same species by different observers in neighbor- 

 ing localities often show highly improbable differences in the appar- 

 ent regularity of arrival. In the records of .the Biological Survey 

 the best example of uniformity in arrival is that of the chimney 

 swift at New Market, Va., as noted by George M. Neese. The dates 

 of each year from 1884 to 1906 are, respectively: April 16, 16, 15, 16, 

 16, 11, 9, 15, 21, 14, 15, 14, 12, 7, 16, 14, 16* 12, 11, 9, 12, 12, 10. 

 The three days, April 14, 15, and 16, include more than half the 

 years, the average date is April 13, and the average variation from 

 this date is only 2.2 days. Usually, however, the recorded dates of 

 arrival of a species vary irregularly from 10 to 14 da}^s, with an 

 average variation of a little more than 3 days* These variations and 



