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



If we imagine that each " wing " (or open half-girdle) of the Pylonida becomes closed 

 by a lattice-work, and so transformed into a hemispherical or roundish cupola, we obtain 

 the characteristic shell of the Tholonida. Indeed every girdle of the former corresponds 

 to a pair of opposite domes of the latter. The axis of each pair of domes is one of the 

 three dimensive axes. 



The primordial chamber of the Tholonida (or the central chamber, around which all 

 cupolas are regularly disposed) is either a simple lentelliiJtical lattice-shell, like 

 Cenolarcus, or it is a trizonal shell iynih. an enclosed concentric medullar}' shell), like 

 Lrxrnacilla. As in both cases the building and the disposition of the cupolas around it 

 are quite the same, we can suppose that the whole family of Tholonida may have been 

 derived originally from Larnacilla (or Trizonium), and that the Cenotholida (with a 

 simple central chamber) are sprung from the Coccotholida (with a ZorHaciZZa-shaped 

 central chamber) by reduction and loss of the original medullary shell. 



The family Tholonida can be divided into three subfamilies according to the disposi- 

 tion of the cupola-pairs in one, two, or three dimensive axes. In the Cubotholida lie 

 two cupolas on the poles of the transverse axis of the central chamber (corresponding to 

 Amphijyylc) ; in the Staurotholida we find four cupolas crosswise disposed, on the poles 

 of the transverse and principal axes (corresponding to Tetrapyle) ; in the Cubotholida are 

 at least six cupolas, on the poles of all three dimensive axes (corresponding to TlioJo- 

 nium). In all three cases the number of cupolas may be augmented by the secondarj- 

 apposition of other chambers or domes in the same disposition. Sometimes also the 

 whole cortical shell becomes enclosed by an external veil or mantle of delicate network. 

 The lentelliptical (or often nearly cubical) central chamber becomes often reduced, so 

 that its sides are incompletely latticed or widely opened ; in some Cubotholida only 

 the twelve edges of the eioht cornered cubical central chamber remain ; its six sides are 

 quite open and only over -vaulted by the six hemispherical cupolas. From the opposite 

 points of the latter (in the deep annular constrictions between them) often arise radial 

 spines, and these lie commonly in diagonal planes, separating the dome-pairs. 



A similar dome-building or a composition of the polythalamous shell by pairs of 

 cupolas we find also in the next (fifth) family, the Zonarida. But here the true cause of 

 the peculiar dome-structure is cpiite difierent, not an apposition of new chambers, but the 

 constriction of a cortical shell-like Larnacal^ns by two or more annular constrictions. 

 These constrictions lie in dimensive j)lanes (or in planes parallel to these), and therefore 

 the cupolas are (aU or partly) in diagonal planes, a condition quite opposite to that found 

 in the Tholonida. One of the annular constrictions is constantly in the sagittal plane 

 (separating the right and left halves of the shell). The number of the constrictions in the 

 few genera is two, three, and four, and therefore the number of the cupolas four, six, or 

 eight. As this cortical shell constantly encloses a trizonal medullary shell (or Larnacilla- 

 shell), we cannot doubt that the Zonarida must be derived from the Larnacida. 



