( 411 ) 
I shall be able to show afterwards, that it is by no means a rare 
case that glands, which might superficially be regarded as water- 
elands, secrete nectar. The agreement between water- and sugar- 
secreting glands is by no means limited to their external shape and 
anatomical structure (see foot-note, p. 402) but extends further. 
It may be remembered that Haberlandt x ) expressed the view that 
there might be a phylogenetic relationship between nectar-glands and 
water-glands. Haberlandt supposes that water-glands are very much 
older and that in many cases nectaries may safely be assumed to 
have been formed from them. He starts from the supposition that 
the difference between water- and nectar-secreting trichomes does not 
so much lie in the difference in composition of the product (for the 
cell-content of water-secreting trichomes also occasionally gives a 
reaction for glucose) as in the dependence on the hydrostatic pressure 
of the water-conduct system in the case of the secretion from water- 
glands, which dependence would not exist in the case of nectaries. 
The gradual evolution of nectaries from hydathodes was regarded 
as giving to the plant the advantage that the secretion became 
independent of the root-pressure. We have already seen in the 
first part of this communication that the basis of this theory is 
incorrect and that it is difficult to defend any longer the hypothesis 
of a phylogenetic connexion between water-glands and nectaries. 
In returning from this digression to observations I must point out 
that the left and right side of the tongue-shaped ray-florets are folded 
over in the bud towards their middle and that they are covered 
with secretory trichomes especially on the veins. 
In the tubular disc-florets they are found especially on the limb 
(Coreopsis tinctoria) or only on the lowest portion of the corollar 
tube (Coreopsis laciniata and Cosmos hybridus). 
Finally I wish to observe that in Dahlia and in Cosmos the corolla 
of the tongue-shaped ray-florets is not only covered with these long, 
glucose-secreting glands, but in addition with short stalked club-like 
glandular hairs. I did not find these in the two species of Coreopsis. 
They are not coloured by dilute stains, whereas the other glands 
^ke up stains. In shape they agree with those, which according to 
Solereder are common among Compositae. The character of the 
walls suggests the secretion of mucilage; probably they only function 
later during the flowering-period. 
From analogy with what may be learnt from Calystegia, Datura 
a nd iVicandra it might be expected that the trichomes which secrete 
x ) Haberlandt, 1. c. p. 58, 
