What causes this triangle effect? (waterfall) I was in a friends garden and saw this:

My question is: what causes the water to flow towards the center? My first thought was that maybe the water in the center falls faster and thus creates a sort of force inwards, but because gravity doesn't care about weight I don't think thats correct...
I also noticed that the width of the water is everywhere the same, exept for the edges. The water which flows towards the center forms a small tube there.
What causes this effect? Or is it simply because of the design of this fountain? (I don't think so, I've seen this before on other designs)
 A: My guess is conservation of mass flow rate. 
Let's imagine to cut the water flow with an imaginary surface: since mass must be conserved, the quantity of mass passing through this surface per unit time must be a constant:
$$\dot m = \rho \ u \ A =  const$$
where $\rho$ is density, $u$ is the velocity of water and $A$ is the cross section of the water flow.
The density is constant in this situation (1). Moreover, let's assume that the thickness of the water flow is almost constant (it looks like it is, up to a good approximation). We will then have 
$$A = const \cdot l$$
where $l$ is the width of the water flow. Therefore
$$u \cdot l = const \rightarrow l = \frac{const}{u}$$
Let's take a $z$ axis to start at the top of the fountain (when the water starts to fall) and to finish when the flow enters in the water; it is then easy to see from conservation of energy that 
$$u (z) = \sqrt{2 g z}$$
so that
$$l(z) = \frac{const}{\sqrt z}$$
(yes, we are assuming that $g$ is also constant: quite reasonable in this case!)
Of course, this formula cannot be completely right, because it would give you a divergence in $z=0$. We are probably neglecting some other kind of energy (I am betting on surface tension), and the "true" form must probably be something like
$$l(z) = \frac{1}{\sqrt {a z + b}} \ \text{cm}$$
where $a$ and $b$ are constants. You can already see by plotting the function $1/\sqrt{z+1}$ that the shape looks similar to the one in the picture. 
So, in conclusion, my opinion is that the shape of the water flow is not that a triangle, but rather that it behaves as $\sim 1/\sqrt z$.
(1) As a matter of fact, the density of water is almost always constant, since water is with very good approximation an incompressible fluid.
A: In regard to the effects on the edges, by the looks of it, it is quite reasonable to think of Kelvin-Helmholtz instabilities, since there will be a shear layer between the water falling at a specific speed and the air being pulled down at a different (probably slower) speed.
