Comparative method, being a great accomplishment of historical linguistics, obviously has its flaws. Reconstructions of proto-languages provided by this method can only be perceived as mere abstract resultant of the material used in comparison.
Have there been any efforts to authenticate these reconstructions in the cases where direct evidence of the mother-language is actually available?
For example, have there been any attempts to reconstruct Proto-Romance disregarding all the evidence from Latin, an then comparing the results to it ?
I am sure that there are more families where such a procedure could be applied.