16 
Siphonophora Rosarum 
of unequal size, the anterior pair being smaller than the posterior: and 
they lie—in close relations with the nervous system—partly in the 
hinder portion of the head, but mainly in the pro-thorax. Viewed from 
above, the larger posterior pair lie just in front of, and above, the first 
pair of thoracic ganglia; while the anterior pair take up a latero-dorsal 
position,—one on each side of the sub-oesophageal ganglia. 
From each gland a duct is given off. The ducts from the anterior 
glands pass backwards, at the same time dipping dowm towards the 
ventral surface. Those from the posterior pair are much shorter, 
and dip down, passing forwards to meet and fuse with those from 
the anterior glands. From this point they continue in their ventral 
course as single ducts, one on each side, closely embracing the 
nervous system. These two ducts eventually meet and join in the 
median ventral line, and then proceed in an anterior direction as a 
single duct, which lies directly beneath the sub-oesophageal ganglion 
(see Plate I, Figs. 17, 18 and 19). This main duct ultimately 
opens near the tip of that part of the labium which forms the small 
under-lip. As has been stated before, S. rosarum has no salivary 
pump to force the secretion of the salivary glands into the wound 
made by the setae, and, as it does not form galls, it is probable 
that the salivary secretion is used to prevent the sap which oozes 
from the wound in the host-plant coagulating when it comes into 
contact with the atmosphere. The way in which the salivary secretion 
is used is probably as follows. The saliva, when the glands are active, 
will be forced out of the aperture in the small under-lip by the pressure 
behind of the newly-formed secretion, and perhaps by some contraction 
of the main salivary duct. It will be remembered that the under-lip is 
closely applied to the base of the beak or proboscis, and the secretion 
being poured out from it would naturally run down the groove which 
traverses the length of the upper surface of the beak. Consequently, 
it would meet and mix with the sap oozing from the wound made by 
the setae, and the mixture of the two would then be sucked up into the 
pharynx and thence into the crop. It is quite possible, therefore, that 
the salivary secretion, besides preventing the coagulation of the sap, 
may contain some digestive ferment to assist in digestion. 
The salivary glands are composed of very large cells—especially the 
posterior pair—with closely granular cell contents and large nuclei with 
well-defined nuclear bodies. 
In the posterior pair, the boundaries of the individual cells separate 
as they pass from the exterior to the interior, leaving spaces between 
