How to calculate threshold value for numeric attributes in Quinlan C4.5 algorithm?

I am trying to find how the C4.5 algorithm determines a threshold value for numeric attributes. I researched and cannot understand, in most places I found this information:

Training samples are first sorted by the values ​​of the Y attribute in question. There is only a finite number of these values, so we denote them in sorted order as {v1, v2, ..., vm}. Any threshold value lying between vi and vi + 1 will have the same effect as dividing cases into those whose attribute value Y lies in {v1, v2, ..., vi} and those whose value is in { vi + 1, vi +2, ..., vm}. Thus, there are only m-1 possible cleavages on Y, all of which must be systematically analyzed to obtain the optimal split.

Usually the midpoint of each interval is selected: (vi + vi + 1) / 2 as a representative threshold. C4.5 selects the threshold vi as the threshold for each interval {vi, vi + 1}, rather than the middle itself.

I am studying the Play / Dont Play example ( table of values ) and don’t understand how you get the number 75 ( tree generated ) for the humidity attribute when the state is sunny, because the humidity values ​​in the sunny state are {70,85,90,95}.

Somebody knows?