Jo IIS. Schmidt 109 



Fig. 3, whei-e hard rays are altogether lacking, exhibits no incurvation 

 at all. 



The number of pigment spots is also, as will be seen from the figures, 

 of physiognomical importance. 



The number of vertebrae has been found to vary from 101 to 126 

 inclusive, a range of variation which is very considerable, and greater 

 than in any other species of fish hitherto investigated. The other 

 characters also have a wide range of variation. Thus we find, for the 

 hard rays — 17, for the pigment spots 7 — 21, and for the pectoral 

 rays 16 — 22. For the sake of brevity I will here restrict myself mainly 

 to the mention of a single character, viz. the number of vertebrae. 



Preliminary investigations soon showed that there was no difference 

 between the sexes in respect of the characters here in question. It was 

 further found, on repeated analyses of the same population, that most of 

 the characters exhibited a high degree of constancy from year to year, 

 as far as the scope of the investigations extends up to now. An instance 

 of such constancy is shown in the graphs of Fig. 4, representing the 

 number of vertebrae in the years 1914, 1915 and 1916 for the same 

 population (St. 31, Ise Fjoi'd, Sealand, Denmark). The number of 

 vertebrae is noted in figures along the vertical lines, each dot denoting 

 a single specimen of Zoarces. It will be seen that the three graphs 

 agree excellently well together, and therewith also the average values 

 for the three years, viz. 113-4:2 (±049), 113-23 (±0-43) and 113-40 

 (±0-44). 



Altogether about 80 population analyses were made, embracing 

 material from the greater part of the area of distribution of Zoarces. 

 The separate population samples consisted as a rule of about 200 

 specimens, which permit of a very good degree of accuracy in the 

 determinations. For vertebrae, the highest mean found was 1 19-44, 

 and the lowest 107-98 ; similar results were arrived at in the case of 

 the other characters, least in the case of the pectoral rays, where there 

 are also, it must be noted, very rarely more than four variates. 



The mean values for the various characters may appear in highly 

 differing combinations in the different populations. Disregarding the 

 pectoral rays, where the range of variation is but slight, and using 

 A, B and C to denote a high value, a, b and c a low mean value for 

 number of vertebrae, hard rays and pigment spots respectively, we find, 

 out of the eight possible combinations, six were realised in our 80 

 population analyses: ABC, ABc, AbC, Abe, abC and abc. Up to the 

 present, however, we have encountered no instance of aBc and aBC, 



