Macintosh, OS X 10.9, Java 7
So. Here's the deal. I've discovered a glitch in where if you place a water stream, 2 high, 2 deep and 1 wide, (that was my testing platform) and then you place skulls on the bottom, then remove the blocks on that row, you can see that the water is frozen as such in it's state (as in, movement state, as such.) If you were to place skulls, and then water, the water would destroy the skulls, however, so order of placement is important in reproducing this.
If you are confused, I'll have a bunch of screenshots and then add description information.
Screenshots uploaded. The first screenshot is to do with the fact that it works on all 5 variations of skulls.
The second screenshot is the setup I was using.
The third screenshot is where I show you I have placed the two heads. Notice they are on the side of a block.
The fourth screenshot is what it looks like once I have removed all the blocks around the heads on that row. Even the ones the heads were on are removed, yet the heads stay and the water does not respond to the fact that it has extra room for movement.
One more side note is that when you give the water an update, it moves again and removes the heads, spreading normally.