inorganic chemistry - Why can't I reproduce this photolytic reaction of ferrocyanide?

I was so intrigued by the reported observation that sunlight precipitates the iron from $\ce{K4Fe(CN)6}$ (originally attributed to Matuschek, 1901) that I wanted to see it myself. I put a saturated aqueous solution, uncovered, in direct sunlight for four hours at mid-day. Absolutely nothing precipitated from the solution!

My understanding from the given reference is that in solution the $\ce{K4Fe(CN)6}$ gives $\ce{Fe(CN)6^{4−}}$ and then:

$\ce{Fe(CN)6^{4−} + 2H2O <-> Fe(CN)5 + (H2O)^{-3} + HCN + OH-}$

The photolysis causes $\ce{Fe(CN)6^{4−} ->[h\nu] Fe(CN)6^{3−} + e-}$, so then we also have

$\ce{Fe(CN)6^{3−} + 2H2O <-> Fe(CN)5 + (H2O)^{-2} + HCN + OH-}$

Is it correct to assume that if a $\ce{Fe(CN)6^{4-}}$ anion absorbs a photon of adequate energy (in this case it appears to require $\lambda < 313nm$) then with 100% probability the anion will photolyze the Fe bond?

And are there models that predict (at least order of magnitude) the absorption probability of a photon by an aqueous anion (being irradiated with a given spectral flux)?

Furthermore, what is the reaction that would precipitate iron in this case? When we bring the K back into the equations it looks like the HCN would react to produce KCN. I don't know what should happen to $\ce{Fe(CN)5}$.

In fact, after several days in open sunlight the solution is subjectively getting darker and I do see a small amount of red sediment at the bottom of the dish that appears to redissolve on swirling. Could be an iron oxide, but would also be consistent with $\ce{K3Fe(CN)6}$. In either case I don't have any indication from the source what should happen to the $\ce{(CN)5}$ groups if iron is precipitating, or what else might take one of the K atoms from the initial compound.

(Ultimately I had hoped to quantify the rate of photolysis by weighing the precipitate.)