[split] Why the VMS text is meaningful or meaningless?

[nickpelling]

I strongly wonder if you're measuring two overlaid (heterokurtotic?) statistics at the same time here.

In that case wouldn't the Central Limit Theorem mean the distribution would look more Gaussian?

You're presuming you know what the two component distributions look like! The whole Gaussian statistical toolbox collapses if you have multiple overlaid distributions, so reason with care here! :-)