of the Fishery Board for Scotland. 



203 



parts as will best express this balance. The skull, for example, is in pro- 

 portion to the caudal region, the height of the body is correlated with the 

 height of the skull, the position of the fins is connected with the move- 

 ments of the body, and so on. The various dimensions which express 

 these relations should be chosen, and not one only of a structure but 

 several. By choosing several dimensions of one structure the conclusions 

 in the end are more certain and satisfactory, even though something is 

 added to the labour. 



The measurement of the mass or weight of the various parts and 

 structures has not yet been attempted on account of the great practical 

 difficulties in the way. When these have been overcome and accuracy in 

 measurement attained a new and valuable field for the study of variations 

 will be opened up. In connection with the taking of the measurements, 

 further, it may be said that each species presents different possibilities. 

 In addition to the general considerations, it is therefore necessary to take 

 note of the characters which differentiate a species from its nearest allies. 

 Systematists have hitherto chosen the superficial characters for the most 

 part in order to distinguish between species, but it is becoming more and 

 more apparent that if we are ever to gain a definite notion of the 

 " species " and its mode of origin in any one case, we must penetrate to 

 the internal morphological differences, and from these limit off the species 

 from one another. The distances between various species (as between 

 herring and sprat, plaice and flounder) may differ in the various genera; 

 we do not yet know. And here again, as Heincke points out, the study 

 of variations methodically carried on is part of the work of the present- 

 day morphologist and systematise 



The methods by which the measurements are taken are various, and 

 depend a great deal on the ingenuity of the observer. Davenport (I.e.) 

 explains the various methods now in use, but there is still scope and need 

 for more accurate ways of taking measurements, This is a separate study 

 in itself demanding special attention, and in a short time it is probable 

 that the laboratory of the biologist will be supplied with instruments as 

 delicate and accurate as those of the physicist and chemist. For the 

 present work the ordinary instruments only have been used, but it is 

 hoped to make some improvements in the future stages of the study. 



When the characters have been measured and their values tabulated it 

 is necessary to reduce them to their percentage-values with respect to some 

 common standard, otherwise they would not be directly comparable. The 

 common standard chosen by different observers is not always the same. 

 The total length of the fish, or this length minus the length of tail, are 

 usually taken. In the present work, the total length minus tail-length and 

 skull-length has been chosen as standard, and is called " body -length " 

 throughout, although the small distance between the tip of the snout and 

 the anterior end of the vomer is included therein. The choice of this 

 standard is not merely a matter of convenience. It has to be chosen so 

 that the differences in the values of the various dimensions are fairly 

 represented in the averages. If the standard dimension is itself under or 

 near 100 of the units employed (millimetres, say), this condition is always 

 fulfilled. But if the standard is large, over 200 and more, then it 

 will be necessary, in order to bring this down to 100, to divide it always 

 by a larger number than 2. Each dimension must be so divided, conse- 

 quently, and the differences between the original values will be reduced 

 according to the value of the divisor. Thus, if the divisor is 3 or 10, then 

 a difference of 3 or 10 in the observed values becomes only 1 in the per- 

 centages. This difficulty is certainly obviated if the standard is taken as 

 1000, and the secondary dimensions expressed in thousandths, but this 

 increases the labour of calculation enormously. It is advisable, therefore, 



N 



