276 MERISTIC VARIATION. [part I. 



scales, (2) the number of the gill-rakers, (3) the alleged difference in 

 the opercula of the two sides. In view of the first point, viz. that the 

 number of the scales on one side is intermediate between that of the 

 Pilchard and that of the Herring, it seemed desirable to know whether 

 the resemblance extended to the minute structure of the scales or was 

 restricted to their number only. On comparing microscopically the 

 scales of the Pilchard and the Herring, I find that those of the 

 Herring bear concentric lines which are almost always smooth and 

 without serrations, while those of the Pilchard are marked with lines 

 which are waved into very characteristic crenelated serrations. On 

 compai'ing the scales which are repeated, it was found that they also 

 shew these characteristic serrations and that in pattern they differ in 

 nowise from the scales of the Pilchard. This evidence appears to tell 

 very strongly against the theory that the small scales are derived from 

 a Herring parent. 



The evidence from the gill-rakers seems to be also unreliable. In 

 a normal Pilchard Mr Day found 71 on the hypo- and cerato-branchials 

 of the first gill-bar, and in a specimen examined by me 72 were present 

 and in normal Herrings 48. But in my two specimens shewing the 

 repeated scales there were present, on the normal sides 79 and 67 

 respectively, and on the abnormal sides 78 in the one fish and 67 in 

 the other. In size and shape the gill-rakers were like those of the 

 Pilchard, being smooth, and unlike those of the Herring, which bear 

 well-marked teeth. 



As it is stated that the serrations characteristic of the operculum 

 of the Pilchard were very distinct on the abnormal side, it is impossible 

 to lay much stress on the circumstance that they were less distinct than 

 those of the other side. 



In addition to the considerations given above, there are several 

 d, priori objections to the hypothesis of the hybrid origin of these 

 forms ; as, for example, that unilateral division of parental characters 

 is certainly not a common phenomenon in hybrids, if it occurs at all, 

 and so on. But since the evidence advanced for the theory of hybrid 

 parentage is already open to criticism, it is perhaps unnecessary to 

 discuss these further difficulties. 



On the whole, therefore, it seems simpler to look on these 

 abnormalities as instances of the phenomenon of Meristic Variation 1 . 



In Ophidia the number of scales occurring in different parts 

 of the body is constant in some genera and species, and variable in 

 others. Variation in the number of rows of scales on the body 

 may be specially referred to as an instance of a change in number 

 occurring at right angles to that just described. The number of 

 such rows in Tropidonotus, for example, is generally 19, but Mr 



404. Boulenger informs me that the Swiss Tropidonotus viperinus 

 has either 21 or else 23 rows. 



405. Tropidonotus natrix is remarkably constant in the posses- 

 sion of 19 rows of body scales. A specimen taken in Switzerland 



1 Compare with an interesting series of cases in Gasterosteus (Stickleback). 

 Boulengee, G. A., Ann. and Mag. N. H., 1893, S. 6, xi. p. 228, see also ZooL, 

 1864, p. 9145; Sauvage, Nouv. Arch, du Mm., 1874; Day, Journ. Linn. Soc, xm. 

 1878, p. 110 ; &c. 



