nlme::summary.lmList can fail when some of the groups are too small for the model being fitted and hence give different-size output vcov matrices.
Over on StackOverflow someone is trying to get this code to work:
(results: Error in `[<-`(`*tmp*`, use, use, ii, value = lst[[ii]]) :
subscript out of bounds)
The problem occurs because one Dog (#9) in the data set only has
data from two days, so the quadratic term is automatically dropped
from the corresponding lm() fit. (See with(Pixel,table(Dog,day)) )
nlme::summary.lmList seems to have some reasonably fancy code that
tries to address this, but I haven't been able to get to the bottom of
the problem yet.