S TIG MARIA. 277 



verse section, and this group is succeeded on the side where it adjoins the 

 inner cortical cylinder by a number of smaller ones with narrow lumina, 

 which form a projecting point. It is highly probable, as Williamson 

 suggests, that this point is the initial group of the bundle. Sometimes a 

 few other isolated broader elements are also found on its other side 1 , which 

 may be explained with Renault to be an indication of an inner wood- 

 portion, a ' bois centripete.' The size of the transverse section of bundles 

 of this type varies extremely, the bundle being often reduced to a few or 

 even to a single one of the broad elements (Fig. 34 C), and the*n the initial 

 group suffers corresponding diminution in size, though it never entirely dis- 

 appears. That the bundles in this, as in the preceding case, are collateral 

 and capable of secondary growth is evident at once from the inspection of a 

 transverse section figured by Williamson 2 , which shows an unmistakable 

 bundle of Stigmaria (Fig 1 . 34 D}\ but joining on to it on the side opposite 

 to the initial group is a well-developed layer of secondary wood with com- 

 paratively narrow elements disposed in regular rows. The bundles of the 

 category which we are here considering are often found in very thin and 

 small appendages ; but it is hardly possible to lay down a rule on this 

 point, for they occur on the other hand in the very broadest, so that apart 

 from the internal structure we cannot very well unite them as states of de- 

 velopment with those of the first-mentioned type. 



A third class again is composed of bundles, which resembling those of 

 the second class in general habit are distinguished from them by the less 

 regular form of the transverse section, which is triangular with the angles 

 rounded off (Fig. 34 E}. But the angle (a) which abuts on the inner 

 cylinder is always more prominent than the other two (b, b}, and in these 

 also we find elements with a smaller transverse section, but varying in num- 

 ber and arrangement, and sometimes combining to fown a narrow band 

 which bounds one side of the bundle. Bundles of this kind have been 

 figured by Williamson 3 especially, and next to him by Renault 4 , who con- 

 siders the three angles to be alike and to be initial strands, and therefore 

 conceives of the entire bundles as triarch root-strands, a view against which 

 Williamson everywhere protests most vigorously. In fact the essential 

 difference between the three angles, one of which only shows indubitable 

 signs of its initial character, and also the excentric position of the strand 

 which never touches the inner cylinder with more than this one angle, must 

 necessarily give rise to very grave doubts. It has already been frequently 

 pointed out, that it is very dangerous to determine the position of the initial 

 strands so directly from the diminution in size of the tracheides. And in 

 this case it is quite possible to take an entirely different view of the matter 



1 Williamson (6), t. 11, f. 59. 2 Williamson (6), t. 11, f. 61. 3 Williamson (6). 



4 Renault (2), vol. i, t. 30, ff. 2, 3, 4, and vol. iii, t. A, ff. i, 2, 4, and (10), tt. i, 2. 



