XVI THE VOYAGE OF H.M.S. CHALLENGER. 



will be grouped at greater or less distances. These two attributes, simplicity and com- 

 pleteness, are evidently appropriate to an ancestral form. To begin with, each part of an 

 organism will, by inheritance, resemble the part from which it has budded out. Succes- 

 sive variations introduce distinctions between the parts of an organism, just as they intro- 

 duce distinctions between one organism and another. At the same time the simplicity 

 sought for must be limited by some standard of completeness, otherwise we should be 

 looking for the origin of things in general, not the ancestry of a particular group. In the 

 structure of an Amphipod we may recognize simplicity in the segments of the perseon, 

 where, as a rule, each segment is to a certain extent free from its neighbour and closely 

 resembles it, and we may recognize it also in the flagella of the antenuse and branches of 

 the pleopods, in which, commonly, numerous joints exhibit one and the same pattern. The 

 theoretical completeness of the appendages rests to some extent on a comparison with 

 other groups of Crustacea, but the limits either of completeness or simplicity which are 

 to be expected in the special group are soon arrived at. If, then, by comparing not only 

 one but every available character in all the families, we at length make some approach to 

 a complete set of ancestral chai'acteristics, we shall be able to construct an ideal Amphipod, 

 with no parts degraded and none exaggerated. And if further, by comparing this ideal 

 with existing species, we find one among them bearing an exceptionally close resemblance 

 to it, such a sj^ecies will have some claim to stand, not perhaps at the head, but in the 

 centre of our classification, as most directly representing the type or original from which 

 the other Amphipoda have in various degrees more widely diverged. As a matter of 

 fact, in the genus Ganimarus the well known species Gammarus pulex and Gammarus 

 locusta are very much of the commonplace facies desired. They are naturally chosen for 

 explanatory purposes and as representative species. They have the requisite completeness; 

 the secondary flagellum of the upper antennae is not wanting as in Am2jhithoe, nor the 

 mandibular palp as in Dexamine and the Orchestidse ; the palp of the first maxillae is 

 not degraded as in Orchestia, nor the maxilliped palp curtailed as in Lafystius ; no 

 segments of the perseon are coalesced as in Dulichia, nor of the pleon as in Atylus and 

 Goplana; the third uropods are not uniramous as in Metopa, nor the second as in 

 Cerapus. They have also the requisite simplicity, as could easily be shown by a 

 detailed comparison with other species. The distribution of these two species lends an 

 additional probability to the view that they represent an ancestral form. Far more than 

 any other Amphipod Gammarus pulex appears to have spread itself over the fresh-watei- 

 streams of the world, and Gammarus pulex is connected by the very closest ties with 

 Gammarus locusta. It is clear from the general distribution of the Gammarina that the 

 chief nurseries whence they issue are the weeds of the coast. From these the rivers are 

 accessible as well as the ocean, yet in the rivers the species of Amphipoda are few, while 

 in the ocean they are multitudinous. This admits of a simple explanation, if we accept 

 Gammarus locusta as representing the ancestral form which at one time occupied the 



