I’m trying to join shapefiles and Census population tabulations at the block level for US states. I’ve noticed that some of the shapefiles have fewer unique GISJOIN values than the tabulations at the same geographic level. I saw on the NHGIS website that water-based census blocks are removed from the shapefiles. However, some of the GISJOINs absent from the shapefiles do not appear to be water-based (i.e., the block ID doesn’t start with 0). I found this to be the case in MA and MI 2010 (I haven’t checked all states). In some states (e.g., Kansas 2010), there were no differences in the unique GISJOIN IDs between the shapefile and the tabulations. Would blocks be excluded from the shapefiles for any reason other than being water, or could a block be water-based without the block ID starting with “0”?