Why does a halocline form? I was doing an experiment over the last couple of days to try to crystallize alum using a thermal gradient. The idea was that solute at the bottom of my container would be dissolved at a higher percentage (I used a coffee mug warmer) because of the solubility curve...
http://nshs-science.net/images/alum_solubility_chart.gif
Then because the local solution is warmer than the solution above it, it would become buoyant, rise to the top, cool, become super saturated, and then deposit its excess solute preferably where I want it. This setup worked fine for me with epsom salt and ordinary table salt. But in this case, using alum, a salinity gradient, or halocline formed and the solution at the bottom, though warm enough to be uncomfortable to the touch, did not become buoyant. My container was only 9 inches tall. At the top, the solution was almost room temperature. I remembered this effect when reading about solar ponds:
http://en.wikipedia.org/wiki/Solar_pond
But typically a solar pond is several feet deep.
My question is, why does a halocline form in the first place? Is this simply the action of gravity on the solute? Is alum's greater solubility the reason I observed this effect in only 9 inches of container height?
 A: This sort of thing happens all the time when you try dissolving a highly-soluble compound in a solvent without proper stirring. A high-concentration, high-density layer forms, and is stabilized by the fact that the density increase due to solute concentration outweighs the density decrease due to the increased temperature. For low-solubility compounds, or compounds with small $dS/dT$ (where $S$ is solubility and $T$ is temperature), this is less of a problem.
In the case of $\text{KAl(SO}_4)_2\cdot12\text{H}_2\text{O}$, the solubility at high temperatures (near boiling) is very high (see Solubilities of Inorganic and Organic Compounds by Atherton Seidell). You mention the solution at the bottom was "warm enough to be uncomfortable to the touch", which is probably in the regime of very high solubility, so it is overwhelmingly likely that this is the mechanism. 
Stir it (magnetically or otherwise), or try using a smaller thermal gradient, and it should probably be better.
Halocline formation

user1839484: My question is, why does a halocline form in the first place?

I'd guess the mechanism for a halocline can roughly be described by considering the density $\rho$ as a function of solute concentration $S$ and temperature $T$, ie, $\rho(S,T)$. The halocline stability condition then becomes
$$\frac{d}{dT}\rho(S,T)>0.$$
Since $S$ depends on $T$, the halocline stability condition becomes
$$\begin{array}{|l|}
\hline
\frac{d}{dT}\rho(S(T),T)=S'(T) \rho ^{(1,0)}(S(T),T)+\rho ^{(0,1)}(S(T),T)>0.
\\\hline
\end{array}$$The second term is almost certainly negative for any value of $S$ (density of solutions generally tend to decrease with increasing temperature). The first term is positive, and is proportional to $S'(T)$, the slope of the solubility curve. If $S'(T)$ is large, the halocline stability condition is satisfied, and dense layers can form.
(Note: I made all of this up, but I think it might be helpful in phenomenological understanding. Also, thank you for teaching me the word "halocline").
