479 
1921.] Bird-Migration by the Marking Method. 
subject, and isolated records must always be regarded with 
suspicion and as at best suggesting the theoretical ex¬ 
planations which they seem to indicate. The chances of a 
faulty record are in themselves almost negligible if the 
method is carefully and scrupulously followed : wrong ring 
numbers have frequently been reported and have been 
speedily detected by being inconsistent with the particulars 
ol* marking of the bird to which the number really belonged. 
There are, however, several records which suggest that the 
individual birds concerned behaved in an abnormal manner 
( cf . Mallard, Section VI.), and this makes it the more 
necessary that all deductions should rest on a broad 
foundation. 
It would obviously be desirable to collect a mass of data 
sufficiently large to be treated statistically, but it cannot 
be said, in view of the numerous unknown factors, that 
this has yet been achieved. In the first place there is 
to be considered the possibility that the material being 
dealt with is not wholly homogeneous : even in the case 
of birds of the same species bred in the same area there 
may be migratory and resident individuals, and therefore 
possibly migratory and resident races. In the case of 
birds caught and marked in winter the material is more 
obviously of mixed origin and may even contain morpho¬ 
logically distinguishable geographical races or subspecies. 
Not only may some individuals of a species be migratory while 
others in the same area are resident, but there is no ground 
for assuming that all the migratory individuals perform 
similar movements : the movements, indeed, certainly differ 
in degree and may differ in kind, and it is not even fair to 
assume that the same individual will act in an identical 
manner in successive years. It follows, also, that great 
caution is necessary in deducing routes of migration from 
records relating to different birds. The obvious temptation 
is to plot on a map all the localities of reappearance and to 
consider them as points in a common path, but it is not 
sound reasoning to say, for instance, that because many 
