Yes, I agree. It is obvious that there is a badly eroded character, unreadable but its ghost is there. There is room for another in the space that has been chipped off, but no telling whether or not there was anything there.

The last visible character looks like a '0', which could be the third character of '1903, with the ghost of '3' following. Without a context, it is not certain that these last characters are indeed the year of death. If they are, then the stone is correct and NIS is in error.