[El Gato]

Gravity causes the liquid in the long end of the tube to fall, which creates a pressure differential along the length of the tube, which atmospheric pressure on the to-be-emptied liquid attempts to "correct".

Doesn't need to be atmospheric pressure, the weight, and thus pressure of the liquid in the tank would be enough.

[agere_contra]

Excellent point El Gato.

The higher end of the siphon would have to be very close to the top of the liquid in a vacuum bell-jar to remove that possibility.