Two distinct spectral references for the same card is not good situation, as each represents a different assertion regarding colorimetric accuracy. All you can tell from a comparison of them is how different from each other they are, not that one is more accurate than the other. Getting at accuracy would require re-measuring the card with a spectrophotometer, and comparing those measurements with the respective reference file measurements. Then, throw out the reference file that is furthest (and that could be a simple Euclidian distance) from the spectrophtotometer measurements.
As I’ve messed with such, having such an instrument has become more compelling. I’ve tried simple wide-band devices such as this:
https://www.sparkfun.com/products/15050
but it won’t offer sufficient resolution for measuring targets. This device:
https://www.amazon.com/X-Rite-EOSTUDIO-i1Studio/dp/B076PN4PMG
would do the trick, but that part of X-Rite spun off into a company called Calibrite, which promply added about $90US to the $489 cost of the device when made by X-Rite. Knew I shoulda bought one last year…