Anyone with an answer to this? It has to be rather complex.
First of all, it cannot be a simple additive/multiplicate combination. If so, there would be a direct corollary between the first and second, the second and third, and the first and third values, but this is not so:
First values in series (12, 16, 20, 24, 28): +4 spacing evenly. Even though each series is not related to the others, if we have a direct/simple combination (using addition or multiplication), we should see a proportional increase in the second and third series. Even if the spacing is not even, it (the spacing) should either increase or decrease regularly.
This is not the case:
second series (21,30,38,54): (+9,+8,+16) - Notice that we have a negative rate of change from the values 1->2(+9) and 2->3(+8), followed by an increase in change from 2->3(+8) to 3->4(+16).
So, if formulaic, it must be a high-order polynomial, modular, sinusoidal, or some other function that allows for this behavior.
If procedural, it could be something to do with prime factors, the number of letter "e"s when the numbers are spelled in German, who knows.
The best I've got is defining the third value in terms of the first (x) is: f(x) = 2x+2^(x/4-2), but it only works for the first three, 4th and 5th need to be 64 and 88 respectively for it to work, and it still does nothing for the missing second-place value.
My guess is that it is some obscure procedural method....