Here's my issue with your logic in coming up with internal fractions: If I'm interpreting it correctly, then that assumes that every horse is running at the winner's speed. Just because the winner is running at a rate of 6.47 lengths/second doesn't mean that a runner in second place making up five lengths will do so at that same rate. He probably is running slower, so that's where it gets tricky, ya' smell me?

Yeah, I've thought of that too but can't really think of any accurate way to correct that.

I dont know what is so hard about this. If the horse is running a 24 sec quarter, it's 7 len. per sec. If a 25 sec. quarter it's about 6 1/2 and a 26 sec quarter about 6 len. Going from memory but I think that's right.

If the horse you're trying to measure is moving much faster than the horse with the known time, then you can adjust it. Like the horse with the know time runs a 25 sec quarter, but your horse made up 5 len, then okay assume he is running a bit faster at: 7 lenghts per sec. Difference is only 0.14 vs 0.15 of a sec/length.

Like ateams example. SInce he made up 5 lengths on the 6 1/2 length/sec horse, just assume he's moving 7 lengths per sec. He made up 5 lengths, multiply by 0.14 and therefore he ran 0.7 sec. faster.

Possible error of .05 sec. doing it the other way? How accurate can you get?

Correct, but not enough to make enough of a difference. Let's say you have a sprint where they went :45 to the half and we looked at 3 horses:
* The leader ran 6.90 lengths/second to that point
* the horse in 2nd by 5 lengths ran 6.78 lengths/second to that point
* the horse in 8th by 12 lengths ran 6.63 lengths/second to that point

So using that information, a fair "one-size-fits-all" lengths beaten adjustment for the pace call would be 6.75 lengths/second.