24 ON THE STUDY OF ZOOGEOGRAPHICAL REGIONS, 



iii. Collection of Records. 



It is essential that the records used ha fairly complete. Pro- 

 bably in very few cases can the complete records of distribution 

 of even a single species be obtainable But this is not necessary, 

 because the object of the method is not to produce contours of 

 impeccable accuracy (such, indeed, are practically an impossi- 

 bility), but to study the type of contour produced. The altera- 

 tion of a contour line a few miles (or even perhaps a few hundred 

 miles) usually will not aftect our ability to recognise it as belong- 

 ino^ to a particular type. As an example of the kind of contour 

 aimed at, one may offer any of the well-known meteorological 

 contour maps drawn over a large area In these, the general 

 distribution of isobars, isohyets, or isothermals, is very clearly 

 shown; but these lines are drawn as free curves, and ignore 

 many small local variations. To give a good example, the 

 average annual rainfall map of Australia (Plate i.) is produced 

 from about seven hundred records. Doubtless, if we could have 

 access to seven thousand records, a much closer approximation to 

 the truth could be obtained. Yet nobody would seriously main- 

 tain that the contour as now produced is not accurate enough for 

 all practical purposes, especially for study as a coinplete ivhole, in 

 which too much attention to detailed curvings of contour-lines 

 would mar the clear effect now obtained. 



Under this heading, it is hoped that the method of Specific 

 Contours will, if adopted, lead to a closer recognition of the value 

 of every single record that can be obtained of every single species^ 

 however common it may he. 



iv. The validity of Species. 



No attempt can here be made to answer the question " What 

 is a species T To each student who desires to use the method, 

 sufficient common sense may be attributed not to mar the result 

 by an insistence on the recognition, as species, of units of lower 

 than specific value. In this connection, it should, how^ever, be 

 clearly noted that, on the whole, both " splitter " and "lumper" 

 will produce approximately the same contours for a given group. 

 For, if a recognised species, A, be subdivided into any number of 



