If you have nothing to split on (the cell is empty) then the line is dropped. If you have a content but no occurence of the separator, the line is kept (= splitted over 1 line).
Using the split option at the read stage, on lines where the splitted column is empty will drop the line altogether. This behaviour is different from the extend.split which keeps lines even when the split content is empty.
That is not true if you split at the read stage. In that case, the line is dropped if you have nothing to split on.
Quick reminder that
U.N
goes starts at 1 and goes up to N.U.N
is in [1..N]