My understanding of siphons is that air pressure is able to lift (because of lower pressure at the top of the siphon) the liquid up the shorter arm and then gravity pulls it down the longer arm.
However, siphons can work in vacuum under special conditions. Wikipedia states that the liquid siphons because of cohesion and tensile strength provided the liquid is pure, degassed and the surface (of the tube) is clean. For example, in this YouTube video, based on this paper (not free), they use a fluid with strong intermolecular attractions.
The pressure explanation tells that the diameter of the tubes is immaterial. However, it seems that if tension is the driving force behind the siphon, then the diameter should have an effect on the working of the siphon simply because with a larger diameter I have more liquid to pull the liquid from the other side.
But increasing the diameter of the shorter arm surely cannot pull the liquid from the lower reservoir to the higher one, right? What effect does the diameter of the tube have, if we stick to the cohesion explanation, in vacuum?
Incidentally, if the liquid consists of long molecules like polyethylene glycol, could it be siphoned in the opposite direction. And if there is a possibility of the liquid being siphoned in the opposite direction, either because of its long molecules or because of high cohesion, can that effect be evened out by the effect of pressure (which pushes it from the higher reservoir to the lower one)?
EDIT: Let me try to make things precise. Let's say we have a top reservoir A, a bottom reservoir B.
- While siphoning from A to B, usually pressure and gravity drive the system. If the liquid under consideration was highly cohesive, can cohesive forces, perhaps in the absence of air pressure, slow down or reverse the flow? What happens if the shorter arm has a large diameter? Here I expect the cohesive forces pulling the liquid in the longer arm to increase, thereby decreasing the flow.
- If we try to siphon from B to A, then under ordinary circumstances, the liquid will just fall back to B, except perhaps a few drops falling into A, after which air fills the pipe. But if the liquid is highly cohesive (or has long molecules), can it be siphoned, perhaps in vacuum, from B to A? In this second setting, gravity will try to drive the flow from A to B, so could it be cancelled by cohesion pulling the liquid from B to A?
It seems impossible that the liquid should be able to go from B to A without someone just sucking everything out of B into A.
As an aside, just a small clarification. Under ordinary circumstances, if A and B were at the same level and I attempted to siphon water, say, from A to B, it will just stay in the pipe without falling on any side, right? Like sucking on a straw and closing the top before the liquid falls.