21st Century Classical Greek -- dependent clause particles

Grammars of Classical Greek define dependent clauses as those starting with one of five particles, and being unable to stand alone to provide meaning.

However, in the very first chapter of Thucydides, a number of clauses cannot stand alone, but do not use these particles. I've talked about this chapter and these clauses before. In this chapter, they use personal gerundives. 

The problem isn't even thinking in Grenglish. In English, we can say, "Thinking it over, I decided I had to take action." You could not use "Thinking it over" stand-alone, because you could also say "they decided..." whatever. 

The problem is copying what your sources say, and your sources copied from theirs, maybe all the way back to Thrax. 

I said that I was looking deeper into this, and I have now gone through Thucydides looking at his usage of the five particles. My conclusion is

a/ they all have multiple purposes.

b/ their main usage is not at the start of dependent clauses but in idioms or common expressions that, unlike idioms, do translate in a fairly straightforward way.

c/ negation can follow all of them, and Thucydides uses both ou and mi to negate them, except for the purpose particle ina, for which he only uses mi.

Dependent clauses are just another case of grammarians hanging their claims on the obvious, like morphology in verbs, and getting things not even wrong sometimes.

Goodwin claims that some of them, like oste, require specific verb forms, like impersonal gerundives and indicative. This is false. Thucydides I 70.9 uses oste to subordinate a conditional using the epistemic. Also, remember that the impersonal gerundive has the nuance of "due and owing". Use with oste might seem a lesser emphasis than the indicative, as in most cases, but "such that an action is due and owing" is a different nuance than "such that an action happened".

And Smyth discusses succession of tenses, which means you have to follow, say, an augmented "tense" with a specific other "tense", but he ignores dependent clauses that don't use the particles.

In a stunning omission, none of the grammars discuss other particles which introduce dependent clauses, like ὥσπερ, which I ran across in Book IV of Thucydides while researching one of the particles they do discuss. This is a case of copying from sources which fail to address all the data. That's a failure of the Test of Occam's Razor.

But context was always the weakest part of the old grammars: they failed to examine every context and that is why LSJ has holes in it and the grammars are not even wrong sometimes.

The only grammar I found that agrees you can start a dependent clause with a personal or impersonal gerundive in Greek, the same as in English, is Kendall Easley's User-Friendly Greek. But it is aimed at the New Testament, which was written in koine Greek not Classical Greek. So what I said about Greek grammars stands. The people writing grammars of Classical Greek failed to do comprehensive work and thus to provide comprehensive -- or even correct -- information.