I would say that the original expression Manilius nesciebat quid scribebat can be read in two ways:
There is an indirect question: He did not know what he was writing about.
There is a relative clause: He did not know the thing about which he was writing.
The answer by brianpck covers the first case, and indeed then one would expect scriberet.
In the second case the indicative mood is perfectly fine.
(Relative clauses can have conjunctive, but it is certainly not required for a simple statement about facts. While not required, I would accept it for emphasis on Manilius' relation to his writings.)
However, the interrogative pronoun must be replaced with a relative pronoun, so one would expect quod instead of quid.
Of the four combinations of these two words only one is invalid, and that is the one that Scaliger picked:
Manilius nesciebat quid scribebat.
Manilius nesciebat quod scribebat.
Manilius nesciebat quid scriberet.
Manilius nesciebat quod scriberet.
It is not easy to tell whether the original intention was an indirect question or a relative clause.
Both make sense, and one could argue that semantically the difference is not all that great.
Either way, there is a grammatical mistake.