Linear practice effect (better performance on each successive treatment, where the amount of improvement from one treatment to the next is always the same).
Then use: Reverse counterbalancing within groups (that is, between subjects - but it is still a within subjects design with regard to the factor).
Six of the 12 rats get A B C D E, the other six get E D C B A.
(The other way to say the same thing is: Within the first group of two rats,
one gets A B C D E and the other gets E D C B A. Then within a second group
you have the same thing. Then within a third group, etc. for six groups.
This is why McBurney calls this "within groups".)
Linear fatigue effects (worse performance on each successive treatment, where the amount of impairment from one treatment to the next is always the same).
Same remedy as for linear practice effects.
Warm up effect (that is, a non linear practice effect) (Relatively poorer performance on the first trial, or trial or two, but then better performance on all the rest).
A single Latin square:
ABCDE
BCDEA
CDEAB
DEABC
EABCD
Two rats run each of the orders. Problem: Two rats left over, or three rats short.
Any time the rat runs the black maze, he may be slower on the maze that follows (sequence effect)
Best remedy is every possible order. Since the number of orders is 5 x 4 x 3 x 2 x 1 = 120, this is not possible. But you could at least randomly select two different latin squares. One rat would run each order. There would still be the problem of two rats left over.