1908-9.] THE STELE OF OsmUNDA CINNAMOMEA. 517 
traces, followed at subsequent nodes by broad or narrow bays of paren- 
chyma, which may extend to the centre of the protostele. It nearly 
always happens that these bays are anticipated by nests of parenchyma 
appearing in the xylem below the level of the leaf-traces. Into the broader 
even of the earliest-formed bays, the phloem dips more or less. The bays 
just described are usually quickly re-filled with xylem, but eventually one 
deeper than any of the preceding is enclosed as the ends of the xylem 
surrounding the gap come together. This incipient stelar pith likewise 
often disappears, and may repeatedly appear and disappear before it is 
finally established as a permanent feature. Permanence is sometimes 
attained at the level of the twentieth node, sometimes higher up on the 
stem. 
The permanent stelar pith at the base of the stem is narrow and com- 
paratively few-celled. Soon after its establishment the overlapping of 
leaf-gaps may begin—a feature not infrequently observable at the level 
of the twentieth to the twenty-fifth node. From this time on, there 
is a slow increase in the size of the pith, and continuously an increase 
in the number of overlapping leaf-gaps. But the pith is yet relatively 
meagre and there are no more than two or three or four overlapping gaps 
when the internal endodermis first appears. It is to be pointed out that 
before, and even after the internal endodermis comes on the scene, indeed, 
even in the adult, leaf-traces may occasionally originate cladostphonically, 
that is from the face of the cylinder of xylem without causing a complete 
break in its continuity. In such cases a pocket of parenchyma is found in 
the cylinder of xylem, which communicates above the leaf-trace with the 
parenchyma surrounding the xylem, but not with the pith. This phe- 
nomenon will be dealt with later on. 
The first indication of the internal endodermis is a single tannini- 
ferous cell marked by the radial ‘‘dots”’ peculiar to the endodermis. ‘This 
is quite rapidly added to until there is a group of three or four or more, and 
finally with the increased numbers, they form a ring within which one and 
then an increasingly larger number of parenchyma cells appear. But the 
endodermal pocket may almost or quite die out and then re-appear before 
a constant internal endodermis is fully established. During the progression 
of these events several leaf-traces take their departure from the stem, 
leaving gaps in the xylem, through which I found no communication of 
internal and external endodermes. One fact, however, may not be 
without significance in this connection, namely, that rows of tannin- 
bearing endodermal cells frequently radiate out from the internal endo- 
dermis into the large leaf-gaps, and similar groups may temporarily occur 
just centrad of the middle line of the departing leaf-traces. In fact, 
