July, 1906 | THE PERCENTAGE OF ERROR IN BIRD MIGRATION RECORDS 
89 
age of error by a combination of many local reports and in the light which they 
seem to throw upon the general nature of migratory arrivals. 
I have, moreover, a peculiar pleasure in presenting some of the results of my 
study on this occasion in as much as the first paper that I ever read before the 
A. O. U., fifteen years ago, dealt with the migration records kept at Philadelphia by 
the seven founders of the Delaware Valley Ornithological Club which had then 
just been organized. 
The plan of recording migration which is now followed by our migration 
corps, as well as the detailed results of the work, have been published each year 
in Cassinia, the annual publication of the Club; but, as many of you probably have 
not had an opportunity of consulting it, I shall give a brief outline of the plan. 
The work is confined to the spring migration, as the study of the fall move- 
ment has proved impracticable except in the case of a very few observers. 
Schedules are mailed in February to all former observers and to probable re- 
cruits, and contain the printed names of 90 migrants, with blanks for the date of 
arrival after each, and a blank column for the entry of additonal species, while re- 
marks and supplementary data are to be written on the back. The convenience 
of having each species in exactly the same position on every schedule can be 
appreciated by any one who has done any compiling from such records. The 
first date of observation only is entered in the arrival column but others may be 
given under remarks, especially when the first arrival was an advance straggler. 
The schedules are returned in June, and as soon as the report for the year is pre- 
pared and published, each observer receives a separate, thus keeping up his inter- 
est in the study and keeping him in touch with his fellow workers. 
From the records of the past four years I have selected the dates of arrival of a 
few of the most common and easily recognized species as reported by some of the 
most reliable observers, that is, those who were most constantly in the field. From 
these we may draw some interesting deductions. 
In the first place the diversity of dates is considerable; even the average of 
arrival for four years is by no means uniform, and one can readily see that any 
calculation on the speed of migration in a general study would vary considerably 
according to which station we should quote as indicating the date of arrival at 
Philadelphia. Indeed the diversity is sometimes as great as that between points 
separated 200 miles or more, as given in some of the published records. [See 
Table I.] 
Tabulating the records in another way and using the whole series we find 
that there are usually scattering reports of arrival from one or two stations. And 
then on one or two days the species reaches nearly all of the other stations. In 
other words the bulk of the arrivals are massed on one or two days. [See Table II.] 
It seems to me that the indications are that early arrivals drop down here and 
there thruout the area covered by our observers, sometimes being first recorded 
from the stations farthest up the river or farthest back on the uplands; then comes 
the bulk movement some days later which marks the advent of the species at all 
the other stations. 
In other cases there is no well marked bulk movement, and the species is re- 
ported arriving day after day at one station or another until it is spread over the 
whole area. In such instances it may be seen regularly by one man some days be- 
fore it appears in the territory covered by his neighbor only a short distance away. 
This method of tabulating our data is probably the most satisfactory, but as an 
illustration of how the combination of several records reduces the percentage of 
error, take sixteen stations within ten miles of Philadelphia in 1903, and we find a 
range of a week, or more, variation in t he reports of arrival of six common species; 
