32 Scott and Brehner . — On the Secondary 
tracheide, the desmogen-strand, after completed division, but 
before elongation of the tracheides, should show in transverse 
section an average number of thirty-one elements only. 
Now this method seems at the first glance very promising; 
as a matter of fact it has led, in the hands of different 
observers, to absolutely contradictory results. Roseler, the 
last author who employed it, obtained, as we have seen, results 
seeming to point to a considerable amount of cell-fusion 1 , a 
conclusion which is quite inconsistent with direct observations 
on serial sections or on macerated material. He found, in 
fact, too many elements in his supposed unaltered desmogen- 
strands ; in the light of other observations we can easily 
explain this by supposing that a certain amount of sliding- 
growth of the tracheides had already taken place. This, how- 
ever, is assuming the point to be proved, and the inefficiency 
of the counting-method by itself is manifest. Apart from the 
obvious difficulty arising from the great variability of the 
number of elements in different bundles (ranging from forty- 
four to ninety-four in the examples counted by us), there is the 
further difficulty that sliding-growth certainly begins before all 
the divisions have taken place, so that we have no fixed point 
at which to begin our numeration. 
Although the counting method does not help us in detail, 
yet the comparison of transverse sections of developing bundles 
is very instructive. We find at a certain stage a small strand, 
with active cell-division in all parts. At a later stage the 
strand is larger, the external portion (in Yucca ) remains un- 
changed, or may show some additional divisions. The more 
internal part, although the seat of the chief growth, shows no 
fresh cell-division whatever. Its increase is due to the in- 
truding ends of young tracheides, growing in from other 
levels. Tracing the changes in older bundles we continue to 
find an enormous increase in the number of elements of the 
xylem of the bundle, as seen in transverse section, without 
any cell-divisions to account for it. This is only to be ex- 
plained by sliding-growth, and in fact affords by itself, a 
1 Roseler, 1 . c. p. 333. 
