The interviews revealed some patients speaking a bit more slowly than usual, though not remarkably so. On the other hand, I found a significant number exhibiting tiny airflows in their speech: Using a metal mirror against the lips to record condensation, I could visually detect a small amount of air coming from their mouths just before they spoke. It was as if they preceded each sentence with a tiny, inaudible sigh.

The more I thought about this, the more intrigued I became. What were these tiny airflows? What function did they serve? I never saw them in the speech of stutterers, nor in the speech of normal speakers with no history of stuttering. Why, then, in this group?

I remember sitting in my office mimicking the flows. And then it hit me. The answer was in the flow. The flow was being used to ensure an opening of the vocal cords prior to speaking. The flow was passive, never pushed, and its sigh-like quality appeared to keep the cords apart and relaxed.

I could now explain the bizarre behavior of a patient whom I had seen earlier who stated that he never stuttered when he smoked. When I asked him to demonstrate, he lit a cigarette, inhaled, let out some of the smoke, and began to speak. The stuttering disappeared.

When I asked him why he thought it helped, he said it relaxed him. I passed it off as a form of psychological distraction. Little did I suspect at that time that the answer to the problem was to be seen in the smoke being blown past my face.

Later, through my reading, I was to discover that the vocal cords are controlled by the breathing centers of the brain. During normal, quiet respiration they open slightly just before inhalation, then close slightly as the air flow is reversed and exhalation begins.

The opening of the vocal cords has been shown to be an active process, the result of the contraction of a single pair of muscles located at the rear of the voice box. The exhalation phase, on the other hand, is passive, the inward movement of the vocal cords occurring as the muscles relax. Research has shown that the most relaxed state of the vocal cords occurs during this expiratory phase of normal, quiet respiration.

In other research, electrodes have been placed on the vocal cords to study the tension patterns associated with speech. The tensions in the vocal cords before speech have also been studied. The research shows that the average person starts to tense his vocal cords between one-third and one-half second before he speaks.

This fraction of a second before speech begins appears to be critical because it is the time during which the vocal cords can and often do lock. I had observed this earlier in my ultrasonic scans of the vocal cords in stutterers. I reasoned that if I could reduce these pre-speech vocal cord tensions I would stand an excellent chance of keeping the cords from locking and thus stop stuttering.

The trick, then, was to somehow learn to exhale just before speaking as if one were not going to speak at all but were simply quietly breathing. The brain had to be fooled into believing that the speaker was simply taking another breath. If the brain were fooled, it would develop no pre-speech tensions on the cords.

I began to experiment with an airflow technique. I started by asking a stutterer to produce a long, audible, relaxed sigh. I then asked him to sigh once again, and when halfway through, say a one-syllable word. The stuttering stopped. I increased the number of one-syllable words spoken on a breath. Again, no stuttering.

I was amazed. It seemed so simple. But day after day my results were confirmed - the patient's speech continued to be fluent. My spirits began to rise. I kept wondering how long before the inevitable relapse, but it did not occur, and I decided to proceed to the next step. I asked the patient to make the flows inaudible. He did this immediately, saying he had practiced this at home since he did not care to go about sounding "like a breathy pervert." His speech was now totally acceptable, and he remained fluent.

Here, I thought, was a treatment that represented the essence of simplicity. It focused specifically upon what I had theorized to be the cause of stuttering. The passive airflow kept the vocal cords apart and relaxed prior to speech and deprived the brain of the signals necessary to trigger the stutter reflex.

But for other patients, the immediacy of the result belied a subsequent period of difficulty. Although they used the airflow continuously, the still stuttered at times. What was wrong?

I recorded examples of these stutter occurrences from a number of patients, I was able to uncover four characteristic mistakes which seemed to provoke the reappearance of stuttering. I called them misuses of air flow, and all patients were subsequently given instructions to alert them to these mistakes so they might avoid them at all times.

1. Pushing the Flow. Under conditions of stress, patients tended to push or force the flow, which frequently led to a locking of the vocal cords and a subsequent stutter. They were reminded that the flow had to be absolutely passive, that a pushed flow would sooner or later be interpreted by the brain as an "h", and that the brain would then "think" that all sentences began with an "h" and the stuttering would reappear.

2. A Failure of Transition. Another source of difficulty was the lack of a smooth flow into the first sound of the first word. The flow appeared short, with a pause between the end of the flow and the beginning of speech. It appeared as if the patient were catching his breath - as if, perhaps to initiate an inhalation. The pause provided the time necessary for the cords to lock. I told patients, "The air flow is your life-line to fluency, follow it directly into the word."

3. A Failure of Intent. Patients are often so preoccupied with the upcoming first sound that their mouths are seen to form the position for that sound during the final minutes of air flow. Thus the air flow is affected by the anticipated sound, and one of the effects of such anticipation is always a tensing of the vocal cords - which often leads to a stutter. Patients are reminded that the object of the air flow is to fool the brain that they are just simply taking another breath. If they start to form the first sound of the first word, the brain is not fooled.

4. Holding the Flow. Patients, in an effort to time their flows, often inhale and hold the air by closing their vocal cords, then start the air flow at the appropriate time by releasing their cords. If they attempt to do this under stress, the hold is often transformed into a lock and they stutter.

Patients were instructed never to hold the flow but to keep the air in continuous movement, that is, to inhale smoothly and then just as smoothly reverse the flow and exhale; there was never to be a stoppage of the flow at any point.

With these cautions in mind my patients were now responding well. The Air Flow Technique, when practiced properly, brought substantial improvement. But occasionally stuttering blocks still occurred. Despite the use of seemingly perfect air flows, some patients continued to experience difficulty.

Reexamining these patients I would, for example, hear a young man use perfect air flow into a "t" and then stutter. I tape-recorded a number of instances of these unexplainable blocks and played them over and over.

I was listening to them at home one evening, wondering if I would ever find a solution, when my wife entered the room commenting, "You know, those people speak awfully fast." And then I realized the obvious fact - they were speed-stressing themselves. If a patient was afraid of a sound, even though he used perfect airflow, he would most certainly rush through the word which contained the sound. As a result of this rushed word, the brain would tense the vocal cord muscles, anticipating rapid speech to follow. It's much the same as when a sprinter tenses his leg muscles a split second before starting a hundred yard dash. But while tensed leg muscles in a sprinter give him a clear advantage, tensed vocal cord muscles in a stutterer obstruct his air flow and result in a stutter.

In a sense, then, getting set to speak should be the same as getting ready to run a twenty-six mile marathon. The marathoner doesn't need the high initial acceleration of a sprinter, and if we measure the tension in his leg muscles a split second before the race, we find considerably less of it.

If our goal with stutterers is to subtract as much tension as possible from the cords prior to the start of speech, a slow start is critical. Indeed, when I forced patients to say the first word slowly, the stuttering aborted.

We have found the combination of the passive air flow and the slowed first word to constitute a powerful defense against stuttering, and it is what I have come to call "perfect technique." With "perfect technique" it is impossible to stutter. I began to inform patients of this very important fact. I arranged multiple-stress speaking situations so that each could see firsthand that this dictum was inevitably true. Patients invariable felt much less anxiety and stress, and the effect of their successful performance was powerful and obvious. Here, for the first time, was a clear-cut prescription for fluency.