Cure III:
Here we start to get into testing with large amounts of raw data. Briefly describing my testing methodology, I started with a control build of a CON using only Cure III w/o Prime Conditioning on with base stats of 234 MND and 399 Healing Magic Skill. I would then adjust these stats carefully with equipment and traits so that one remained the same while increasing the other. Trials were conducted by just self-healing myself. Since in 1.19 you cure for max amount even if you're already at max HP, it made this less painful to test. Results as follows:
Control: 234 MND, 399 Healing Magic Skill
n = 324, criticals = 28 (8.64%)
mean = 854.1, median = 854, minimum = 829, maximum = 880
c.mean = 1042, c.median = 1040, c.min = 1010, c.max = 1069
Healing(+) Build: 234 MND, 411 Healing Magic Skill (+12 skill from control)
n = 177, criticals = 16 (9.04%)
mean = 872.0, median = 871, minimum = 845, maximum = 898
c.mean = 1062.7, c.median = 1062, c.min = 1039, c.max = 1090
Mind(+) Build: 280 MND, 411 Healing Magic Skill (+46 MND from healing build)
n = 107, criticals = 12 (11.21%)
mean = 881.5, median = 881, minimum = 856, maximum = 909
c.mean = 1062.7, c.median = 1063, c.min = 1043, c.max = 1083
These tests had a total n of 608, which is fairly limited, but I felt comfortable because at the conclusion of each test, the mean, median, and average of min and max values were extremely close to one another. Unfortunately, it is difficult a formula with this data. If we were to make the huge assumption that the formula works similarly to XI, then you could get this:
Cure III = [ 196 ] + { [Healing Magic] x 1.5 } + { [ MND ] x 0.25 }
However, there is no indication other than speculation that both Healing Magic and/or MND follow a direct modifier increase (as opposed to say multiplicative), so I would take the above formula with a grain of salt. I would also take the critical hit data with a grain of salt as well due to limited trials, other than the fact that when you critical hit, there seems to be a rough +20-25% HP cured increase - that much I'm confident in.
Lastly, with regards to the distribution of cure values on the control (n=324), I found that the distribution did not follow any sort of "normal distribution", and felt it was random more than anything. Below is a graph which plots the frequency of each value between the min and max seen in the control test. Had there been some kind of normal or bimodal distribution, I think I would have seem something by 327 trials.
http://img717.imageshack.us/img717/4076/asdfec.png
* Healing Magic +1 improves Cure III more than MND +1
* I'm unwilling to go any further than that regarding an accurate formula
* When critical hit occurs, the HP cured increases by roughly 20-25%
* Cures appear to be a random value between a set min and max value (not normal)
* No testing was done on Second Wind I/II, Cure I/II or the Sacrifice Spells
The above testing falls far short of being able to find a definitive formula for Cure III; however, I think it does tell us a couple of important things. First off, that the randomness in cure values does not seem to follow a normal distribution, but a random one (where all values between min and max have an equal chance of occuring). Secondly, it at least gives us the general idea that 1 or 1, healing magic is better than MND. I gave a sample formula, which is likely horribly flawed, but you're free to follow up on it if you'd like. In that sample, I imply that every Healing magic skill increases Cure III by roughly 1.5 HP while every MND increases it by 0.25 HP. Finally, Crits seem to add 20-25% HP cured, though I have not tested anything regarding Crit gear. Overall, I'd say a Healing Magic build on Head/Weapon may be worthwhile for a dedicated healer, but givent he expense of Healer's Hand IV, I probably wouldn't invest in any double socketing at this point. Max MP builds (or MND since it adds MP as well) are probably a more realistic goal for a dedicated healer in 1.19. Again, something well worth revisiting in 1.20 with stat allocation.