272 FAMILY NUMMULINIDA. 



Plate XVIII, fi?. 8 ; but such exceptions undoubtedly do occur,* although it is more common 

 to find the columns stopping short after they have been traced through several floors, or to 

 find that, without an absolute loss of continuity, their axes are shifted a little to one side or the 

 other — of both of which conditions examples are seen in fig. 9. Extensions of the non-tubular 

 substance from the septal bands of the spiral lamina over its intermediate substance are not 

 uncommon ; and it is stated by jMr. Carter (xxiii a, p. 321) that in T. jjerforata these 

 extensions inosculate so as to form a minute reticulation all over the cameral spaces. 



464. The intcrseptal portion of the '" canal-system" of Nuiiij/ii/liita bears a very close 

 general resemblance to that of Opcrcidiiw. In every septum (Plate XVIII, fig. 7,,/) there seem 

 to be at least two principal branches (seen in horizontal section at/',/, fig. 4, and in vertical 

 section at f ,/', fig- 8) which originate from the marginal cord of the previous convolution, and 

 which converge as they pass outwards, and inosculate so as to form a network (fig. 7), some 

 branches from which enter the marginal cord of tiieir own convolution as shown in fig. 6. 

 The marginal cord, however, somewhat difl'ers in structure from that of Opcrculliia. Although 

 such a minute reticulation as that shown in the siliceous cast of i\". stria1a,\ represented in 

 Plate XXII, fig. 0, may undoubtedly exist, yet the evidence of sections all tends to show 

 that the marginal cord is usually penetrated by canals of a larger size, of which some run 

 across it either radially or obliquely, w^iilst others traverse it in a longitudinal direction, the 

 two sets occasionally inosculating, but not forming anything like a network. The radiating 

 canals are well seen in vertical sections which divide the cord transversely (Plate XVIII, 

 fig. 2, d , a) ; and their existence was pointed out by me in my former memoir, although I 

 was not then aware of their import. Such sections also show the orifices of the large longi- 

 tudinal canals which they divide transversely. The disposition of these canals is somewhat 

 irregular ; but generally there are found two passing in close proximity to each other along 

 the inner edge of the cord, of which one is often brought into view in sections traversing the 



* I cannot but feel .surprised at the protest entered by MM. D'Arcbiac and Haime (p. 48), 

 against the accuracv of tlic figure in my former memoir on Nummulilcs of which fig. 8, Plate 

 XVIII, is a copv. Not only are tliose figures exact representations of a specimen in my possession, 

 draivn bv an artist (Mr. George West), whose accuracy in tlie delineation of subjects of this class 

 cannot (I ventiu-e to say) be surpassed, but they are precisely paralleled as to this particular by several 

 of the admirable figures in !MM. D'Arcbiac and Haime's own Monograph ; whicli show an equal 

 continuitv in the non-tubular columns (sec especially plate iii, fig. 7, e, and plate iv, fig. 9, d). 

 The theoretical objection drawn from the spiral plan of construction seems to me altogether without 

 weight, wheu it is borne in mind how entirely destitute of any definite direction are the alar pro- 

 longations of the septa in Nummulites of the sinuate and reticulate types. I am strongly inclined to 

 believe, from considerations based on the probable mode of formation of the tubular shells (^1f 59, 442), 

 that the disposition of the alar prolongations of the septa will be governed in no small degree by that 

 of the tubular and solid portions of the surface of the spiral lamina of the previous convolution, over 

 which they extend themselves. 



t It is worthy of note that this species belongs to the radiate group, which is most nearly allied 

 to Operculina. 



