If your design includes a random factor, model the random factor; period.

Exploit the fact that you know a blocking factor exists, and include it into the model. Just because it has small ish variance doesn mean it should be ignored. Exchange the word wet with painted and it works.