y22 Beer. — S Indies in Spore Development . II. 
nuclear phenomena of the Compositae, in so far as these are to be seen in 
the pollen mother-cells and (for comparison) in certain somatic cells of these 
plants. I have reserved all details regarding the structure of the pollen- 
membrane for a separate account, but I should like, before I conclude this 
article, to add one word concerning a general feature of the pollen-wall of 
the Compositae. 
Among all the species which I have examined for this purpose I have 
found constantly two different types of pollen-wall. 1 In the one type the 
exine is flat and unfolded, whilst in the other type it is thrown into a num- 
ber of distinct folds. These two types appear to be constant for the two 
great divisions of the Compositae. All the members of the Tubuliflorae 
which I have examined possess pollen-grains with an unfolded exine (Figs. 
74, 76), whilst all the Liguliflorae which I have observed exhibited pollen- 
grains with a folded exospore (Fig. 75). 
Summary. 
1. During the period just preceding synapsis the nuclei of those Com- 
positae which have been examined were found to contain a more or less 
fine reticulum. No definite aggregations of chromatin which could be 
regarded as prochromosomes were in any case found upon this reticulum. 
2. The reticulum passes into the synaptic contraction, during which it 
is gradually converted into a very long, delicate spireme. No chromatic 
aggregates were observed during this stage, and no definite parallelism of 
the threads occurs other than is inevitable in any closely coiled system 
of filaments. 
3. The spireme which emerges from the closely wound synaptic knot 
has become thicker, and, in favourable examples, shows a series of chromatic 
particles (chromomeres) embedded in a less deeply stained linen thread. 
The spireme, during the hollow spireme stage, either remains unsplit or may 
be very obscurely divided longitudinally. No evidence was found of the 
existence of two parallel spiremes. 
4. The spireme again draws itself together in the £ second contraction 
In doing so it forms a series of loops, radiating from a common centre. 
The loops become detached, thicken considerably, and pass towards the 
periphery of the nucleus. Here they assume the shape of rings, crosses, 
parallel rods, twisted rods, or remain as loops. 
There can be no doubt that the loops which are formed in this way are 
actually the bivalent chromosomes of the heterotype division, and that they 
are constituted of two univalent chromosomes joined end to end. 
1 I have examined up to the present the following species : Tragopogon pratensis , Matricaria 
Chamomilla , Doronicum plant agineum, Crepis taraxacifolia, C. virens, Arctium Lappa , Olearia 
Haastii , Sonchus asper , S. oleraceus, Cnicus arvensis , Hieracium Pilosella, Tagetes plumilla , 
Gaillardia sp., Coreopsis sp., Hypochacris radicata, Beilis perennis , Senecio vulgaris , Anthemis 
Cotula , Artemisia vulgaris , Tussilago Farfara , Petasites fragrans , Taraxacum vulgare. 
