How to calculate t-values from percent ranks?

A table from a psychological test is giving percent ranks (PR).

How can I transform those into t-values by calculation? (Assuming normal distribution.)

I found a lot of references on how to transform z --> t, but none on PR --> t.

Way back in university, we had tables with the bell-curve, to look up PR, z-values, t-values and stanines - but how to calculate?

-
check this one en.wikipedia.org/wiki/Student's_t-test – Seyhmus Güngören Feb 22 '13 at 22:43
When you say $t$-values, it suggests that you're using one Gosset's $t$-distributions (not the normal distribution, which always uses $z$-values). You need the $\textit{inverse}$ of the cumulative distribution function (CDF) for the particular distribution. – Sammy Black Feb 22 '13 at 22:44
@SeyhmusGüngören: I think this is different from what I need. – michi Feb 22 '13 at 22:50
@SammyBlack: I am not very experienced in statistics, but I am quite sure that those psychological tests at least assume normal distribution. You're right, z-values are common, but t-values, too. I found a lot of references on how to transform z --> t, but none on PR --> t. Way back in university, we had tables with the bell-curve, PR, z-values, t-values and stanines - but how to calculate? – michi Feb 22 '13 at 22:54
– cactus314 Feb 22 '13 at 23:26