THE THALIACEA. 93 



cases occurred small specimens, some 400 in number, which had not begun or had not 

 completed their degeneration, and therefore showed their characteristic structure : they 

 were demonstrably of a type hitherto uodescribed. The remaining blastozooids 

 (excepting four, possibly five, specimens to be dealt with later) appeared to pass by easy 

 gradation into this type, though altered by the usual metamorphosis and degeneration. 

 In every case in which phorozooids of Tritonis were captured this blastozooid occurred. 



The presumption is that this und escribed blastozooid belongs to one of the species 

 already captured in the North-eastern Atlantic, namely, to denticulatum, Nationalis, 

 Krohni, Tritonis, or sp., Borgert *. Of these, de^iticulatum is known only as a warm- 

 water form, and has not been taken north of 42° N. ; D. sp., Borgert, is apparently a 

 deep-water form, and I did not take it above 250-150 fathoms; the blastozooid of X^Wiwt 

 has been well figured and described, and does not agree with the one before xis ; there 

 remain Nationalis and Tritonis. As between these two, the presumption is enormously 

 in favour of the one to which belong 97 per cent, of the gonozooids and all the 

 phorozooids. In default, therefore, of characters which enable older blastozooids to be 

 specifically differentiated, I refer all blastozooids from the epiplankton (except the five 

 already referred to) to 



Blastozooid A (? = Dolioltjm Tritonis, Herdman). 



The table (p. 92) and the drawing (PI. 8. fig. 1) sufficiently show the diagnostic 

 characters. 



In tte case of the Thaliacea, as in that of the Chsetognatha, I have endeavoured to sort out every 

 specimen in the collection. That this is practically impossible, especially when dealing with minute 

 specimens, I know well ; even if every particle of the catch were sorted out into its respective Order 

 (which has not in every case been done), there would remain, in rich hauls from 100 fathoms particularly, 

 specimens tangled together in clots from which they can only be freed by tearing them. In such hauls 

 the error in minute specimens may perhaps be as much as 10 per cent.f; in the poorer hauls it drops 

 proportionately, till it becomes zero. But I believe the approximation to accuracy to be close enough 

 to allow one to handle the resulting statistics of population (compare Methods and Data, p. 7, supra), 

 and in any case the conclusions based on counted specimens can only be put forward provisionally for 

 confirmation or refutation by future observers. 



The hauls used for statistics may be regarded as fairly comparable, except for the varying rate of tlie 

 ship's drift : for this factor no proper correction can be made. Hauls made with a small-mouthed net 

 of 180 meshes per linear inch have been neglected, as in the case of Chietognatha, because the catching- 

 power was obviously so slight (simultaneously, 180 meshes caught none, 60 meshes caught G8 specimens ; 

 180 meshes caught none, 60 meshes caught 10 specimens). Hauls with a net of about 18 meshes per 

 linear inch, and a diagonal of the mesh-aperture about 1*5 mm., have also been omitted (but not in tiic 

 case of the larger Chaetognatha), as allowing small specimens to slip through. 



In the course of the second day's work (July 8) we struck a " swarm " of DoUoliim, a 

 remarkable feature of which was that, unlike the swarms of gonozooids hitherto described, 



* Vide Chart, PI. 9. 



t ilr. J. J. Lister, testing this in Ilastiijeritia, found in rich hauls an error in my sorting of 5 to G per cent. 



