FREUD says somewhere that dreams are "guardians of sleep," mediating wishes or worries that might otherwise awaken us. But one ancient kind of dream, far from guarding sleep, devastates it, jarring us from the arms of Morpheus into terrified wakefulness: the nightmare.

Surprisingly, the ancient adversary of sleep has been little studied or even speculated about. For instance, I count only two passing references to nightmares in Freud's The Interpretation of Dreams, his magnum opus.

Ernest Hartmann's book is the latest of his pioneering inquiries into sleep and sleep disorders, although one whose conclusions about nightmares are in some ways as unsettling as the phenomenon iself. Nightmares, he finds, are far more serious, intractable problems than we imagine.

In psychological studies of this sort the method is a large part of the message. Hartmann accordingly devotes a substantial part of this book to the story of how he recruited his clinical subjects, differentiated them as "pure" nightmare sufferers (at least one a week over an extended period) from sufferers with related sleep disorders, such as night terrors and post-traumatic nightmares. He tells us about his control groups and how he used them (for instance, he studied two groups of Vietnam veterans with nightmare problems, for contrast to these "civilians"); and he describes his testing (Rorschach, "locus of control," multiphasic personality, and others) and what it showed.

From Hartmann's sample, a number of broad conclusions about nightmares and their victims emerged. The chronic nightmare victim is likely to have been afflicted for a long time, ofen from early childhood. He or she is often separated by one to three years from a younger sibling, not unusual in itself; but the birth of the younger sibling is likely to have been a disruptive event. The chronic nightmare victim is likely to have experienced, from childhood, a feeling of "specialness" and unusual sensitivity, and often had a stormy adolescence.

IN THE GROUP examined by Hartmann and his associates many had been in psychotherapy; an unusual number had gone through psychotic episodes; some had attempted suicide or thought about it; and almost all had had drug or alcohol problems in the past. The profile established by various tests showed, moreover, an unusually high predisposition to symptoms that could clearly be diagnosed as psychotic, or borderline, in the Diagnostic and Statistical Manual III of the American Psychiatric Association. In none of the control groups, including the victims of traumatic nightmares induced by shocking one- time experiences, were these clear DS-III diagnoses possible.

Hartmann was driven to conclude that chronic nightmares may be closely associated with schizophrenia and other grave disorders; and indeed many of his subjects had family members who'd been mentally ill. (As in vulnerability to schizophrenia, some genetic connection may be suspected).

On a more positive note, chronic nightmare sufferers tended to be free of "neurotic" symptoms (such as phobias and anxieties of the kind asssociated with insomnia). They were often cordial, creative, candid people who talked freely and volubly about themselves -- "open," in a word. Their openness, however, was the positive side of a deeper vulnerability, and their defenses "primitive." Undoubtedly the most interesting and original conclusion of the Hartmann study is that chronic or "pure" nightmare victims, along with children and creative artists, have what he calls "thin" or "permeable" ego boundaries. They tend, that is not to distinguish as most of us between various states -- between fantasy and reality, sleep and wakefulness (their waking up process often lasts a half an hour to an hour), their own personalities and those of others. Their personal relationships typically swing between intense "fusion" and alienation. Even geographical boundaries are, for such personalities, less sharply drawn: Many of Hartmann's subjects had been crime victims because they seemed not to know a "dangerous" neighborhood or time of day or night from a safe one. In all senses, nightmare victims stand at the opposite pole from "obsessional" personalities, who tend to draw rigid boundaries and isolate their emotions to the point of inaccessibility.

The affinity between nightmare victims and children and artists, Hartmann suggests, derives from their sharing "thin" or weak personal boundaries. Artists must be capable of a childlike access to primitive impulses and perceptions. That is essential to their roles as "mine canaries of civilization." Unfortunately, few sufferers from chronic nightmares quite have the talent, the drive or self-discipline to channel their creative vulnerabilities so constructively. When they succeed in doing so, however, their nightmares often subside and their sense of helplessness grows far less acute.

Thus the sufferer.

As for the nightmare itself, it is in form a deeper, longer, more vivid and terrifying extension of the dream life. In clinical studies by Hartmann and others, sleep has been found to revolve at regular intervals during the night between dreamless, relatively tranquil periods of so-called "S-sleep," and five or six intervals of "D-" or "rem" (rapid eye movement) sleep, when dreams occur. Nightmares are likely to come during a rem- sleep phase, late in the sleeping period (one way of distinguishing them from night terrors, which usually occur early). The associated body and brain chemistry and activity are in striking ways closer to wakefulness than during "S" sleep.

NIGHTMARES vary in content, yet hew to common recurrent themes -- the fear or sensation of being chased, threatened, attacked, multilated; and this is precisely the dream substance of small children suffering from nightmares. In normal cases regular nightmares stop or become far less frequent with maturation. Not so among chronic sufferers. Hartmann writes: "The cast of characters in the nightmare world does 'grow up' a bit: 'fewer monsters and tigers . . . more thugs, gangs and armies. However, the most frightening nightmares in adults seem to relate to the same basic childhood fears."

In short, the chronic nightmare victim may be said to pay a great price for the almost childlike openness and receptiveness that flows from his or her "thin" and "permeable" ego boundaries. In Hartmann's subjects, Rorschach tests showed, as if in confirmation, some genuine fuzziness and confusion between "reality" and "fantasy."

What causes nightmares? Stress, for one thing, seems usually to be involved in their onset. But most "pure" sufferers suffer for years, many lifelong. Hartmann is appropriately tentative, but he cautiously theorizes that nightmares "may be related to a lack of support by the mother in the first two years of life . . . " This is obviously a rather orthodox psychoanalytic point of view, but Hartmann is far from glibly or conventionally psychoanalytic. He adds, by the way, some interesting speculations on the "biology" of nightmares. From the tentative findings available, they're associated with an imbalance of neurotransmitters in the brain, and with a disturbance of the brain stem. Hence, presumably, a gestalt theory would find it appropriate that these terrifying dreams of a primitive and childish content would spring, indirectly, from that segment of the brain that links us to the primitive pre-humanoid past.

Much of the final section of this fascinating book is given over to clinical suggestions for fellow psychotherapists, with illustrative case-study examples. Often the best therapy is self-administered -- getting well organized to achieve balance in the waking life: a good marriage or stable friendships, for instance, or creative achievement. Sometimes drug regimens are helpful. The key clinical conclusion, however, is that chronic nightmares may herald serious psychosis and should not, therefore, be taken lightly.

The Nightmare is a wonderfully stimulating, informative book about a subject of universal interest.