Operator error is always a possibility, as is mere coincidence, post hoc ergo propter hoc. These are complicated machines, after all, and I often wonder that they work so well as they do, but it was still an interesting occurrence. Here are some more facts. The carts that were low were 110 ml starter carts, so when they reported back 10%, we are probably talking about ~10 ml in reserve, not 30 or 70 ml as might be the case with larger carts. My routine is to do a nozzle check and head cleaning before printing any patch charts. On the whole, I've been very lucky and have experienced few clogs anyway. After replacing a cart, I always turn off the printer once it runs its priming cycle, pull the plug, wait a bit, then restart the printer, do a nozzle check and head cleaning if needed. My intent is to re-initialize everything, and that may have fixed the problem itself, whatever it may have been.
I agree that there is a large amount of ink in the feeder tubes, but I wonder if there is a shortage of ink in the cart, would it effect the pressure in the line. This might not make much difference when printing a tiny spot on a 8" patch chart, but when the head is swinging back and forth through 24" printing in each direction, the demand on the ink supply must be much greater. I am just guessing, but to me it's an interesting suggestion.
I can say that with the Canon 6300, and before that the 6200, I do run the carts until they are dry, which sometimes happens in the middle of a print. And with the Canon, I have never noticed any color errors due to low ink, before it just stops, that is.