musing on https://math.stackexchange.com/questions/3760125/on-the-value-of-a-skew-schu... and playing a bit, I was happy to find a fast implementation of the principal specialisation of the Schur function in https://dspace.mit.edu/bitstream/handle/1721.1/107965/26_2016_Article_312.pd.... Isn’t the web great? Using that, I looked at what would happen to the high degree polynomial in q if I further specialised it. For prime n, say n=11, set q to exp(2 i pi/n) or, equivalently, transform q^k to q^mod(k,n). For \lambda = {4,3,3,1} this transforms q^12 + 3 q^13 + 8 q^14 + ... + 8 q^96 +3 q^97 + q^98 to 679536 (1 + q + q^2 + ... + q^10) Generally, for n prime, we get as coefficient : c(\lambda , n) /n where c is the content of \lambda in n variables. Here, ‘generally’ means ‘for all partitions with an empty n-core’ alias ‘all partitions that are not hook-shaped’. For these cases only, the coefficient is increased by (-1)^l /n with l being the length of the partition and an extra –(-1)^l is added (to the constant term). The fun was mainly in figuring out how to identify and correct these ‘abnormal’ cases. That’s ‘playing’. The real math would of course start with trying to see the how and why of it, aka as ‘proving it’. Apologies to those who dislike partitions into more than 1 part. Wouter.