134 I'HK entomologist's rkcokd. 



outlines, so that it stands out more boldly. In the male it has more 

 extent in that the spot on disc of forewing is invariably present and 

 often larger and in that a little streak, or a series of tiny dots, homo- 

 logous to this spot, often becomes very distinct next to the costa, just 

 within the apical crescent; the corresponding streak on hindwing also 

 becomes prominent in both sexes, and in the female a spot sometimes 

 appears between the third median and the first cubital nervure, homo- 

 logous to the anterior spot of forewing. 



Whilst in the first generation the nervural streaks on the underside 

 of hindwing vary comparatively little in extent, in the two following 

 this part of the pattern acquires primary importance and affords some 

 of the leading features of geographical variation. In those races in 

 which the second and the third generation differing less from the first on 

 account of the extent of their pattern, still strongly resembling that of 

 the latter, there is an interesting phenomenon to observe of variation 

 along two collateral lines. It is evidently connected with the phenome- 

 non I have dealt with in my introductory remarks to the grades of the 

 first brood, which in the Arctic and more particularly in the Alpine 

 forms may exhibit an intensive dark pattern on upperside of some 

 females and a very limited one on their underside. The generation we 

 are now dealing with develops this feature to a high degree individually, 

 and it becomes also so constant in some regions as to constitute an im- 

 portant feature of the race. We thus have one line of variation in 

 which the streaks of underside keep about as extensive as in the first 

 generation, whilst the upperside pattern becomes more and more 

 characteristic of the two other generations by losing the nervural 

 one and developing the true pattern, and another line of variation in 

 which the underside streaks tend to obliteration, whilst the upperside 

 ones keep quite broad, especially on the outer part of the wing, and 

 even the ground colour may preserve its brj/diiiac-Wke bright yellow 

 colour. The existence of these two lines constitutes a difficulty when 

 one sets to work to establish the various grades of extent of the pattern, 

 because by the upperside some races should stand at the head of the 

 series, whereas by the underside they should be classified further down. 

 What one wants is a classification which will show as nearly as possible 

 the relationship of the various races, and not an artificial one based 

 only on one or on a few characters, chosen according to our fancy, such 

 as used to be the fashion in old days. If we follow this rule, and we 

 take into account both sexes and both surfaces of wings, we find that 

 the races of luipi fall into a very natural order, and that it is only a 

 minority of individuals which seem to be out of place by some charac- 

 ters, due chiefly to the dimorphic tendency I have described in the first 

 brood of hri/oiiiac, airtica, and interjccta. This tendency, so conspicuous 

 in the first grades, is carried on through all the others, and we thus 

 always find individuals with an upperside pattern exceptionally exten- 

 sive for the grade and as compared with that of the underside. 



Grade I. is so very similar to the first generation of some races, and 

 more precisely to loiwris, Vrty., from the marshes of Tuscany and from 

 the Alps, that one could not make out it was a second generation, 

 except by knowing when and where it was collected. I take as typical 

 of this grade and name linnaei, a series collected at Norrweken, in 

 Central Sweden, from July 25th to the beginning of August. I 



