their knowledge of saltpetre from the barbarians; we are no longer left wondering whence the Greeks got their saltpetre, and why they gave the name of “sea-fire” to a mixture in no way connected with the sea; and we are no longer perplexed by the fact that the earliest recipes for Greek fire contain no saltpetre.[87] It remains to inquire how the sea-fire was expelled from the siphons. There were two kinds of siphons, large siphons and hand-siphons. Of the hand-siphons there were several patterns. Some seem to have been thrown by hand, like squibs;[88] from others, mentioned by Cameniata, the charge was projected by air[89]—presumably by a bellows or some such contrivance; while in a third kind, described by the Princess Anna, a pellet was blown by the breath through a flame placed before the front end of the tube.[90] The two latter siphons were of the same species, and as Anna’s was charged with Greek fire[91] we may suppose Cameniata’s took a similar charge. The large sea-fire siphon was fixed in the bow of the ship and served by the two foremost rowers, one of whom laid the siphon and was called the siphonator, while the other, we may suppose, loaded it. The siphon was mounted on a swivel, as may be gathered from the account given by the Princess Anna of the naval battle fought near the island of Rhodes in 1103 by the Greeks and the Pisans. The latter were terrified, she says, by an apparatus which directed on them fire of an extraordinary nature. “Ordinary flame rises upwards, but this flame shot downwards and sideways as well, at the will of the gunner.”[92] Unless the siphon was mounted on a swivel, the phrase which I have translated by “at the will of the gunner” (ἐφ’ ἃ βούλεται ὁ πέμπων) would be meaningless.