668 R. C. PUNNETT. 
I have therefore estimated the variability of three separate regions of the body in Pt. 
laccadivensis, viz. the branchial region, the genito-hepatic region, reaching from the hind end 
of the branchial region to the point of junction of light and dark hepatic caeca, and the 
hepatic-caudal, extending from this point of junction to the anus. In the following table 
this is expressed quantitatively when the unit of variation’ selected was the same in all 
three cases, 7.e. collar length x 1. 


TABLE 10. 
| | ayy ox 100 
| | c | CV.=- Ste 
| 
| Branchial TESION' |. 7. .csse + ‘6712 | 29°66 
Genito-hepatic region il +1°015 | 20°72 
Hepatic-caudal region ...| +2°711 11-79 



From this table it appears that the variability as measured by the standard deviation 
gradually rises as we recede from the collar. But these three regions are of very different 
length and the unit of measurement (collar length) bears towards the branchial region a 
relation as regards length very different to that which it bears towards the hepatic-caudal 
region. To avoid this difficulty we must consider the coefficient of variation (C.V.) in each 
case, and when this is worked out (Table 10) it is at once evident that the relative variability 
of the branchial region as regards length is far greater than that of the genito-hepatic, 
and this again than that of the hepatic-caudal. In a word, as we proceed caudalwards from 
the collar the relative variability of the body as regards length becomes less. In Amphiowus 
and the Vertebrata on the other hand the relative variability of the body as measured by 
the number of segments becomes greater as we proceed caudalwards. One of three alter- 
natives would seem to follow:—either (1) we must regard the processes of differential growth 
as widely different in the two cases, or (2) we must look upon the number of a meristic 
series as altogether independent of its length, or (8) we must consider that there exists in 
Ptychodera a somewhat vague segmentation—that the animal is composed of morphologically 
equivalent though ill-defined segments of which the length differs considerably in different 
regions of the body. The amount of favour with which we can regard the first of these 
suggestions depends upon our conception of the relations of the Enteropneusts to the Chordata. 
Those who are inclined to uphold the chordate affinities of this group would naturally turn 
to either of the other two alternatives. The former of these I regard as untenable, since 
such data as I have been able to collect on the subject all tend to shew that there is 
a fairly high correlation between the relative total length of a vertebrate axial meristic 
series and the number of units of which it is composed. There remains to be considered 
the last of these hypotheses. Are there grounds for attributing any form of segmentation® 
to the Enteropneustic body, and further, if this is the case, for considering that such 
segments may differ in size in the different regions of the animal? To the first part of 
the question the answer must be in the affirmative. In the branchiae, in the hepatic caeca, 
and in the epidermal annulations, we have series of repeated parts of which the latter two 
1 The choice of the unit is often a matter of convenience. 2 Apart of course from the primitive segmentation into 
Thus in Text-fig. 121, p. 666, the unit of variation selected is _ the three regions, proboscis, collar, and trunk. 
collar length x 4. 
