848 
MR. W. GARDINER ON TPIE CONTINUITY OF THE 
In concluding the subject of Palm endosperms I might make a few remarks upon 
some particular examples which appear to be of equal interest. 
In all the seeds nearly related to Calamus the structure is very typical (Plate 69, 
fig. 24). I have already noticed the great development of the middle lamella in these 
endosperms. Another interesting fact is, that in Calamus and Metroxylon a well- 
marked cubical crystal is present, imbedded in the wall of each cell. It seems as if 
there had been a period in the life of the cell when the protoplasm had required to 
get rid of some of the calcium oxalate resulting from the metabolic activity of the 
protoplasm. This was consequently thrown down in the form of a crystal which 
adhered to the cell wall, and in the subsequent thickening which occurred, was 
gradually covered in until it was at length surrounded on all sides by the cellulose. 
Locloicea sechellarum is of interest, not only as affording one of the clearest 
examples of the perforation both of the wall and pit membrane, but also because of 
its very unique distribution (Plate 69, fig. 19). 
In Oncosperma the threads are excessively fine, and certainly suggest the extreme 
probability of the existence of threads which are so delicate as to be invisible (Plate 
69, fig. 21). In fact, I am inclined to believe that this really is so in such endosperms 
as Cocos nucifera. To this seed Martinezia caryctifolia presents a useful transition. 
With iodine it can be seen that very fine threads do go through the almost smooth 
walls, but upon treatment with picric-HoFMANN’s-blue the individual threads cannot 
be distinguished, and only a blue coloration occurs. In Cocos, which has essentially 
the same structure, I was unable to observe threads, though I cannot doubt that such 
threads do exist. In all the Cocoinece the walls are thin and must be carefully 
examined. They are, however, of extreme value, both from the point of view of 
analogy and comparison. In Syagrus (Plate 69, fig. 25) and Desmoncus the threads 
are well seen. Heterospathe data is a particularly favourable endosperm for 
demonstrating the perforation of the middle lamella, which here is but little 
developed (Plate 69, figs. 22 and 23). The threads appear very clearly with iodine. 
In Phytelephas, although the walls are extremely thick the pits are small, the pit 
membrane somewhat thin, and the threads are demonstrated with difficulty. The 
cells contain but little remains of the protoplasm, and several results have induced 
me to think that the amount of solid matter in the perforating thread channels is 
so small that the channels are practically empty (Plate 70, fig. 26). 
These results seem to show that in all the Palmse the structure of the endosperm 
cells is similar. 
Endosperms other than those of Palms .—As I have remarked elsewhere, the 
endosperm of Palm seeds is particularly favourable material for an investigation of 
the perforation of the cell wall by protoplasmic filaments. And, speaking generally, 
in the examination of most endosperms other than those of the Palmce, additional 
difficulties are presented which gi’eatly interfere with successful observation. 
Especially does it become apparent that the thickness of the pit membrane is not 
