JAMES W. CLARKE has noted the curiosity that anyone who shoots a prominent public figure in America -- he deals here with 16 assassins and would-be assassins, most of whom assailed presidents or presidential candidates -- is for that reason alone usually written off as a lunatic. The act of assassination or attempted assassination per se triggers and even shapes a reductionist response, glibly psychological: He did it; he must be crazy.

Not so, Clarke insists. It is hardly difficult, by detailed reference to recent and earlier history, to show that it is often a mistake to assimilate assassins to a single psychoanalytic stereotype. Indeed, it is strange that Clarke is the first to question this conventional wisdom which, as he shows, has filtered into most of the official and quasi-official literature.

American Assassins is an attempt to retrieve the lost or undiscovered distinctions among these manifestly antisocial persons--to show that they obeyed promptings as various, and often as politically and socially intelligible, as the rest of us: if, to be sure, more violently and reprehensibly.

In Clarke's scheme -- professors must have their categories, of course -- there are four broad types of assassins: They range from John Wilkes Booth among the definitely "political" assassins, who was not the least bit insane by any useful definition, to Charles W. Guiteau, the murderer of President James Garfield, who definitely was. (Guiteau later pleaded "divine pressure on me," but his brother John had "no doubt that masturbation and self-abuse is at the bottom of his . . . mental imbecility.")

In between, Clarke finds two intermediate types: the "compensatory" assassin who is trying to remedy a lack in himself (Lee Harvey Oswald), and the "nihilist" assassin (Arthur Bremer) whose overmastering rage at society, or his lot in life, becomes narrowly focused on some public person.

Clarke argues that, aside from the outright crazies who suffer from delusion, most assassins are far from psychotic. He also makes the familiar but valid point that a glib application of psychiatric or pseudo-psychiatric "explanations" to one and all drains clinical terms of useful meaning.

We must, of course, stipulate that any violent assailant of a public figure is at least strange; beyond that, however, we must content ourselves with a greater effort to distinguish and discriminate. Motives, experiences and objectives -- where the latter are clear -- differ. It is senseless, for instance, to lump together the two Puerto Rican nationalists who opened fire at Blair House one autumn day in 1949 and John W. Hinckley Jr., who was trying to get the attention of an actress. So far, then, Clarke makes a convincing case.

Seldom, however, has so sound a conception been less adroitly handled than it is, in general, throughout Clarke's book.

Clarke is a clumsy writer (his trademark is the wandering modifier, as in: "Eligible for parole again in 1984" -- he is speaking here of Sirhan Sirhan -- "the Arab-American Realtions Committee is now working for his release. . . " or -- of Giuseppe Zangara, who attacked Franklin D. Roosevelt in Miami in 1933 -- "At the age of six, after just beginning school, Giuseppe's father put him to work").

More seriously, Clarke has given his imagination free play with the "objective" circumstances surrounding assassination episodes. His laudable aim is to recreate for the reader the plausibility (from the assassin's perspective) of his act of rage, revenge, anger, or political symbolism.

But Clarke, alas, lacks the rhetorical skill to bring off these ambitious technical exercises. Instead of imaginative insight, we too often get passages like this description of Arthur Bremer's mother: "The very real difficulties she experienced as a youth (sic) produced a suspicious, withdrawn woman who was given to erratic hostile outbursts at . . . her children."

One can only admire Clarke's brave attempt to empathize, as he seeks to grasp and elaborate the motives and background of the ultimate antisocial act. But it doesn't work. In most cases, it draws Clarke into a glibness about causation that, paradoxically, seems little improvement on the psychological reductions he deplores.

What, for instance, is gained by rescuing an Oswald (or a Samuel Byck, who in 1974 tried to hijack an airliner, intending to crash it into the White House) from glib psychologizing, only to tell us that "each misplaced occuptional frustration and projected spousal rejection onto political surrogates. . ."?

So far as one knows, neither Oswald nor Byck--the one killed within days of his crime, the other shot and killed on the scene--was examined by anyone competent to make such a judgment about motivation, even tentatively.

Perhaps we must leave to geniuses the effort to penetrate the minds and souls of assassins, while the rest of us make do with the fragmentary documented facts and clear inferences that may be drawn from them.

This is the irony of Clarke's ambitious book, which is otherwise full of interesting information and arresting speculation. Assassins are different, not only from you and me but from one another. Not all are crazy--unless you think it crazy to resent a bad lot or to will one's death for a cause (as, incidentally, millions have done in warfare and other violent enterprises throughout history). Beyond that, however, a gloomy mystery remains. Clarke, for all his earnestness, has not seen far, or sharply, into the gloom.