Dependent Subjunctives seem to have developed from the three original uses of the independent subjunctive. Thus, two sentences like "He is to go." (A jussive subjunctive.) and "His father says so." (an indicative statement) were joined to become "His father says he is to go." The first example is called paratactic, or "co-ordinate." The second is called hypotactic, or "subordinate." Any clause that was originally in the subjunctive as a co-ordinate (paratactic) expression will keep the subjunctive when it becomes a subordinate (hypotactic) clause. As the Latin language changed the subjunctive mood became used more and more as a mark of subordination, even in clauses that could not have been subjunctive clauses when the language was more paratactic. Hence, the name of the mood, "subjunctive" or the mood that subjoins clauses to the main clause. In Classical Latin, however, most subordinate clauses with the subjunctive are still clear developments of their origin in paratactic or co-ordinate clauses.
The jussive subjunctive is used to give orders; that is, it gives direct expression to someone's will, purpose, goal and so on. The most obvious subordinate use of the jussive subjunctive is the indirect command. One imagines two original statements: taceant! dux imperat. = "Let them be quiet! The leader orders it." These two sentences can become a single sentence by subordinating the command itself to the verb or commanding. In Latin you may use the conjunction ut or not: dux imperat (ut) taceant = "the leader orders that they be quiet." Just as you would use ne for a negative command, so you use ne for the conjunction introducing a negative indirect command: dux imperat ne dicant. = "The leader orders that they not talk." When the clauses are perceived as subordinate clauses, they are considered the direct object of the verb of ordering. In other words, they are jussive noun clauses. Any verb of persuading, cajoling, resolving, deciding, ordering, advising, seeking, trying, and so on can take an indirect command. Purpose clauses: Subordinate jussive subjunctives may also act as jussive adverb clauses; that is, they may modify the main verb by indicating "how" someone acted, under what terms or provisions. Such clauses require the use of ut or ne, and they indicate the intention (that is, the indirect command) in the mind of the subject of the controlling verb. Consider: "He did this (thinking) let's help them." In Latin: hoc fecit ut eos adiuvaret. Purpose Clauses, then, are just ways of drawing the actor's will or intention (in the form on an indirect command) into the sentence as an adverbial modifier of the action. There are lots of other ways to express purpose, but this is the most common and handy. Final relative clauses or the Relative Clause of Purpose. A final clause describes the goal or end (Latin: finis) of an action. Typically this purpose or intention is an adverbial modification of the verb (In Grammar as in Law, intention modifies action.). But there are other was to append a subordinate clause to a main clause, and a very useful one is the relative. The relative joins a clause to a noun: "the man who spoke to Caesar." Now, in Latin, if you wanted to say "I sent them (thinking) let them speak to Caesar.", the clause "let them speak to Caesar" would be a jussive clause expressing the purpose in the mind of the sender. I could join this clause to the main sentence as an adverb clause (ut...) or as a relative clause modifying "them": misi eos qui Caesarem adloqueretur. Such constructions are handy and common with verbs of sending and verbs of motion. When the subjunctive is found with relative clauses of place-where it is often of the jussive type: habes locum ubi sedeas. = "you have a place to sit" or "... where you should sit." The Proviso Clause is an special development of the jussive subjunctive: introduced by dum, dum modo or just modo the jussive subjunctive means "Only (modo) Let them do this the while (dum)." Thus, the famous slogan of the tyrant, oderint dum (modo) metuant means "Let them hate [me]; (only) let them fear me the while."
The potential subjunctive is used to make statements in which the speaker represents an event as something he or she has imagined or thought about as likely or possible. The fact that the speaker represents an event as something he or she has thought about no doubt contributed to the rule in Classical Latin that all subordinate clauses in indirect discourse take the subjunctive: it was the mood used for reporting things not as events but as things thought about, and so as things spoken about by another. Result Clauses. When an event is reported and the speaker follows that report with his or her own reflection on what is likely to happen, the potential is called a "potential result." "He is so stupid! He would do it." Under some circumstances, it is easy to infer that the likely event actually happend; in this case we have an actual result or an historical result. Historically, it seems that these two clauses (the event and its potential) coalesced into a single sentence with a main clause (the event) followed by a subordinate clause (the result). This kind of subordination (called hypotaxis) produces result clauses like: tam stultus est ut hoc faciat = "he is so stupid he would do it." or "He is so stupid that he would do it." and tam stultus erat ut hoc faceret = "he was so stupid that he did do that." When the historians felt it necessary to distinguish explicitly the potential result from the actual result, they used the perfect subjunctive. Otherwise, rules of sequence apply. Result clauses are also called consecutive clauses because they tell you what is likely (potential) to follow (consequor) an initial action. Result clauses are often adverbial clauses: "How stupid is he? Stupid enough to do it." After verbs that can take results as their direct objects (efficit = "he did it," "he produced [a result]") or as their subjects (fit = "it happens," "[the result] happens", result clauses may be noun clauses. There may also be relative clauses of result: nemo est qui hoc faciat. = "There is no one who would do this." Relative clauses of characteristic. The important thing about relative clauses of characteristic is that the subjunctive indicates that the person, thing or event is being imagined or thought about in terms of its likelihood, its potential, its characteristics -- not just in terms of its facticity. If you are talking about real people, the subjunctive indicates that you are talking about the kind of people they are, their potential as well as the actual fact. On the other hand, you might use the subjunctive to imagine a whole class of people, regardless of how many actually exist: "There are present men who ...." Such a statement calls attention both to the fact that there are such people and to the tenor of the time, place or occasion that allows such people to exist. Finally, if you are just imagining a kind of person who does not exist, you must use the subjunctive. When the antecedent does not exist (when it is expressed as a negative), the relative clause MUST be in the subjunctive: Its just an example of Roman literalness. If there is no one who would do something, then they cannot be real people about whom we make reports; they must be imagined people, potential events. So, you must use the subjunctive. Since the subjunctive designates potential, it asks the reader to be engaged in thinking about the potential of a person, thing, or event. You can infer cause or result or even concession from potential; but it is important to remember that such clauses are causal or concessive only by inference. We can do the same thing in English, but with a different mechanism.
The optative subjunctive is used to give direct expression to prayers, fears, curses, and so on. Fearing Clauses. The most common context in which one issues prayers, etc., is when one is afraid: "May he come! I'm afraid." Now, if you think about this, the only reason to pray that someone come is that you are afraid that he won't come. This means that the paratactic pair of sentences above is rightly intepreted as "I'm afraid that he won't come." Well, that is just what happened in Latin. vereor ut veniat.= "I'm afraid that he won't come." Similarly, vereor ne veneat = "I'm afraid that he's coming" or [paratacticly] "May he not come! I'm afraid." or "I'm afraid lest he come." The use of ut to mean, essentially, ne non = "that" was apparently confusing at times to the Romans themselves. As a result, from about the time of Cicero on, Romans preferred to say vereor ne = "I'm afraid lest.." and vereor ne non ... = "I'm afraid that not ..."