124 
INTERCELLULAR RELATIONS OF PROTOPLASTS. 
me to agree with Russow that in roots the sieve pores are not 
closed in winter. As the two winters during which I have 
from time to time investigated this point, viz., 1882-3 and 
1883-4, have been remarkable for their mildness, I cannot 
venture to say what might be the case in seasons when 
the cold is intense; nor have I examined any half- 
hardy plants to see whether their behaviour is different. 
In the case of stems the closure in winter has not been 
.—' 
nearly so complete as previous statements would lead 
to believe. Here again, however, the extreme mildness 
of the two seasons may have had some effect. The general 
view of Janczewski is that the function of the callus is to 
close the sieve pores in winter. Russow has, however, pointed 
out the frequency with which in winter the callus shows 
differentiation, in the form of strias passing from cell to cell. 
These striae are probably mucilaginous filaments of proto¬ 
plasmic material; the swelling callus has closed upon them, 
but has not shut them out. The perforations are, therefore, 
probably not completely closed. 
When open the sieve pores are lined by a callus layer, 
which continues into the main callus deposits on either side 
of the sieve plate. When the pores are closed in winter the 
closure operates by means of the swelling of the callus in 
autumn. Once closed the pores remain so through the 
winter. In spring the callus in each pore contracts and the 
pore again becomes visible. This reopening can be produced 
artificially, e.g., by passing a branch of a vine for a week or 
so into a warm moist chamber. The actual data of the 
reopening no doubt varies from year to year, as well as from 
plant to plant. On January 20tli, 1883, after about three 
weeks of very open weather, I found (at Bonn) the sieve 
tubes of Syringa vulgaris (the Lilac) fully open. 
The relations of the constituent elements of the sieve 
tubes in different great groups of plants afford a field of 
evolutionary speculation. In all the investigated Vascular 
Cryptogams the sieve tubes are present, but they are always 
closed—never communicate by’ pores. The membranes 
forming their end walls are pitted, but are usually unpro¬ 
vided with callus. In Pteris aquilina (the Bracken Fern), which 
one would not naturally suspect of a high degree of develop¬ 
ment, the walls are perforate at the pits, but the perforations 
are completely closed by the callus formation ; so that here 
the sieve tubes appear to show the initial stage of a higher 
development. 
In Grymnosperms, on the other hand, the sieve plates, both 
terminal and radial, are callous when young, prior to the 
