The Eyes: the HGN Test, Part 1 of 5

It’s one in the morning and you’re standing on the shoulder of a road.

There’s a pen about a foot from your face. A man you met four minutes ago is telling you to hold your head still and follow it with your eyes only. So you do that. Left, right, left, right. It takes about a minute.

Then he has you walk a line. Then he has you stand on one leg. And then he decides.

Here’s the thing about that minute with the pen. You couldn’t see what he saw. Neither can a jury. There’s no video of your eyes, because a camera on a man’s chest at one in the morning cannot resolve what he says he was looking at. There’s no number. There’s no printout. Nobody else was watching.

There is a man who will testify about what your eyes did, and there is no way on earth to check him.

That sounds like the end of the conversation. It’s actually the beginning of one, and it took me a lot of training to understand why.

I’m Rory Safir, and this is Reasonably Safir. Today it’s just the one subject. Horizontal gaze nystagmus. The strangest piece of evidence in a Florida DUI case, and the only one where the result is genuinely unreviewable.

Stay with me, because the punchline is that it doesn’t matter.

Let me tell you where I’m coming from first, because it’s the reason I can say any of this.

I’m a criminal defense lawyer in St. Petersburg. I was an assistant public defender before that. And at some point I decided that if I was going to fight this evidence for a living, I should go learn it from the people who teach it.

So I took the NHTSA practitioner course. Twenty four hours, three days, and it’s the course that qualifies officers to give these tests. Then I took the thirty two hour instructor course I hold a certificate for, which is the one that qualifies a person to teach it to them. I took both through Impaired Driving Specialists, under an instructor trainer who has trained thousands of officers. Then sixteen hours of ARIDE, the advanced roadside impaired driving course. And then another sixteen, a drug recognition expert and trial techniques seminar.

I did not take that last one to become a drug recognition expert. I took it to be able to take one apart.

So when I tell you what happens in that classroom, I’m not describing something I read about.

And I want to be honest about what that training did and did not give me. It did not give me a way to see your client’s eyes. Nobody has that. What it gave me is the other half.

You can’t see the result. You can see everything else.

Every second of that pen is on video with a timestamp. And the procedure is written down to the inch and to the second. And every document I’m going to use against this test today is one the State already owns. Their manual, their studies, their own reference guide. I’m not bringing in an expert to fight theirs. I’m reading them their own book. Which means the one thing in this case that nobody can review sits inside a procedure that anybody can.

That’s the whole episode. This is part one of five. Let’s go.

Start with the word, because almost nobody explains it and it isn’t complicated.

Nystagmus is an involuntary jerking of the eye. That’s what every officer says on every witness stand in this state. It’s true, and it tells you nothing.

Here’s what’s actually happening.

Your eye is holding a target off to the side. Not straight ahead. Off at the edge of where it can go. And holding a position out there is work. Past a certain amount of central nervous system depression, the eye can’t hold it, so it drifts. Slowly. Back toward the middle, which is where an eye wants to sit anyway.

And then the nerves catch up and snap it back out to the target.

Slow drift off. Quick snap back. Over and over.

That’s the jerk. That’s the whole thing. That’s what he’s scoring.

And once you can picture that, you understand the single most important fact about this evidence. A slow drift and a fast recovery is the only motion an eye can make in that situation. There is no second way to do it.

So alcohol looks like an inner ear problem. Which looks like an old head injury. Which looks like multiple sclerosis. Which looks like an eye tracking a passing train, or the overhead lights on a patrol car cycling behind the officer’s shoulder.

All of it looks the same. Not similar. The same. Because the eye has no other option.

The man at the roadside is not a doctor. He didn’t take a medical history. He cannot tell alcohol nystagmus from the other kinds by looking at it.

And that’s not my opinion. NHTSA’s own training says nystagmus may be caused by pathological disorders, and it names brain tumors, other brain damage, and diseases of the inner ear. The ARIDE curriculum tells officers there are over forty different types of nystagmus, and then says that in this training we’ll be focusing on two.

Over forty. Focus on two.

Now here’s the piece of physiology I think is the strongest thing in the whole subject, and I almost never hear anybody use it.

Alcohol is systemic. You drink it, it goes into your blood, it circulates, and it reaches the whole brain at once. It cannot depress the control of one eye and leave the other one alone. Whatever alcohol is doing to your eyes, it has to be doing to both of them.

And the test is built on exactly that. Three clues in each eye. Six in all. A true alcohol picture is an even one.

So when the officer writes down five, he’s written down something the manual says alcohol usually doesn’t do. And here’s where it gets interesting, because the State has two different explanations for that, sitting in two different books, pointing opposite ways.

Hold onto that. There’s a trap in it and I’ll come back for it.

All right. The test itself.

The stimulus, which is a pen or a penlight, goes twelve to fifteen inches from the nose, and slightly above eye level.

I want to tell you why slightly above, because I’ve watched a lot of lawyers cross this test and I have almost never heard anybody ask.

It isn’t about the muscles. It’s held high so the eyelids pull back and the whole eyeball is visible. That’s the entire reason. Which means a stimulus held too high is an administration error. Not a catastrophe, and I’m not going to tell you it manufactures a clue, because NHTSA’s own research doesn’t say that. What it is, is a departure from a written instruction, on video, that you can measure.

Before he counts anything, he’s supposed to run three checks. Are the pupils the same size. Is there nystagmus at rest, with the eyes just looking straight ahead. Do the eyes track together.

Those exist to catch a medical problem before anybody starts scoring. And the manual says, in its own words, that this observation is a medical assessment. If the pupils are noticeably unequal, it tells the officer the person may have a prosthetic eye, a head injury, or a neurological disorder.

Then the current manual adds something almost no officer has ever been asked about.

If there’s anything abnormal on those pre checks, he may choose not to continue. And if he continues anyway, the manual says that does not follow the standardized protocol, and he should say so in his report.

Let me say that again. His own training tells him that going forward after a bad pre check is a departure, and that he has to document it.

Go look at his report. I’ve never seen one that does.

One number while we’re here, so you don’t oversell this. Something like thirty eight percent of normal healthy people have a small difference in pupil size at any given moment, up to about six tenths of a millimeter. That’s from the prosecutors’ own guide on this test. So unequal pupils by themselves are common.

What you’re after isn’t that the pupils differed. It’s whether he looked.

Then three clues, in order.

Lack of smooth pursuit. He moves the pen from the center out to the side, and the eye is supposed to follow it the way a marble rolls across glass. If it can’t, and instead catches up in little jerks, that’s clue one. About two seconds out. About two back. At least two full passes, each eye checked at least twice.

Distinct and sustained nystagmus at maximum deviation. He takes the eye all the way out to the corner and holds it there and watches. Minimum four seconds. Each eye checked twice.

And onset of nystagmus prior to forty five degrees. He moves the pen out slowly, about four seconds or more, watching for where the jerking starts. Before forty five, that’s clue three.

Four or more out of six and he’s got what he came for.

Now. Slow down here, because this is where sitting in that classroom actually pays.

Go back to clue two. Minimum four seconds at the corner.

Left eye. Right eye. Then repeat the pair, because each eye gets checked twice.

That’s four separate holds. Four seconds each, minimum.

Sixteen seconds of that pen sitting still out at the corners before the maximum deviation check is finished on both eyes. That’s the floor. Not a target, a floor.

And I want to be careful about why that number is different from the others, because the difference is what makes it usable.

The two second pass and the four second sweep are prescribed rates. If he goes faster, that’s a departure from the standardized administration, and it’s an argument you have to stand up and make.

The four second hold is a stated minimum. Four holds at a four second minimum cannot happen in less than sixteen seconds.

That isn’t an argument. That’s arithmetic. And a stopwatch settles it.

Every body camera in America is already running that clock.

So why does the hold exist at all? His own side answers that, and the answer is the case. The four seconds are there to make sure the officer didn’t cause the jerking himself. Because if you take an eye out to the far corner and it jerks for a second and then settles down, that’s end point nystagmus, and stone cold sober people do it every day of their lives.

And here’s a sentence that is still sitting in the current manual.

If you think you see only slight nystagmus at this stage of the test, or if you have to convince yourself nystagmus is present, then it isn’t really there.

Say that to a jury. Then turn to the officer and ask him whether he had to convince himself.

Now the part that opens the whole thing up.

Two of the three clues can be manufactured by the officer’s own hand.

Lack of smooth pursuit is the easy one. Move the pen too fast and a sober eye can’t smoothly chase it. It’ll track in little catch up jerks, because that’s what eyes do when the target outruns them.

Now, I want to be careful here, because this is where a lot of us overreach. NHTSA studied exactly this. They ran the pass faster than the standard and looked at what happened. And what they found is that a fast pass makes officers miss the clue, not invent it.

So I’m not going to stand here and tell you a quick pen built the clue against your client. What I’ll tell you is that he departed from a rate the manual prints, on video, and that the agency that wrote the manual has published research saying the departure changes what the officer sees. That’s enough. It’s better than enough, because it’s theirs.

Distinct nystagmus at maximum deviation can be built in both directions. Hold too short and you’re counting the end point jerk that the four seconds exist to screen out. Hold too long and you fatigue the eye until it jerks on its own. The manual even gives you the far boundary, which is that fatigue nystagmus comes on when the eyes are held out at the corner for at least thirty seconds.

So the window is real, and he’s running it in the dark with no timer.

Only onset before forty five degrees resists, and only partly. It’s the one clue the officer’s own hand can’t directly manufacture, because it depends on where the eye is when the jerking starts rather than on how he moved the pen. Which is exactly why the next part matters so much.

Which means a six clue score is only worth what the administration behind it was worth.

And that reframes everything. The question stops being what did the eyes do. It becomes what did the officer do to the eyes.

Which brings me to forty five degrees. And this one is worse than it looks.

How does he measure the angle?

He doesn’t. There’s no device. The manual tells him it isn’t difficult to determine but it takes some practice, and then gives him a rule of thumb. If you started twelve to fifteen inches out in front of the nose, you’re at forty five degrees when you’ve moved the same distance to the side.

And then it says that at forty five degrees, some white will usually still be visible in the corner of the eye. For most people.

And then, in the very next sentence, it concedes that some people’s eyes may not show white in the corner at forty five degrees at all.

So the only field proxy he has fails for an unstated share of the population, and the manual knows it.

There’s a practice template in the back of the book. A square with a diagonal line. And the manual says, in writing, that the template is for practice only and is not to be used with actual subjects.

Now hold that up next to how this was originally done.

The 1977 research protocol that produced this test used an apparatus. There was a chin rest. The subject covered one eye. The examiner moved the light out to a measured thirty degrees, stopped, and observed. Then out to a measured forty degrees and observed again.

And the protocol said, in writing, that jerking observed at a gaze more extreme than forty five degrees should not be relied upon as an index of intoxication.

Chin rest. One eye covered. Measured stops.

None of that survived onto the side of the road.

And in the 1981 validation study, they trained the officers on the angle by covering the markings on the measuring device and having them estimate against it, over and over, with immediate feedback, until they could hit it within three degrees three times in a row. The researchers brought in one subject whose eyes would only go to about forty three degrees, and another whose eyes went to sixty five. The report doesn’t say why. I read that as making sure nobody could pass by guessing, and you can decide whether that reading is fair.

Three degrees. Verified against an instrument. That’s what a validation officer had to do.

Nothing in the twenty four hour course your officer took requires any of that.

So that’s the test as it was designed, and the test as it gets given, and they are not the same test.

Here’s your question, for your motion to suppress, or for cross at trial.

Ask him how he was trained to find forty five degrees. Not whether he can find it. How he was trained to.

Then ask what he was checked against when he learned it. In the research, the answer was three degrees, verified against an instrument, three times in a row. Ask what his was.

Next time, the number everybody quotes. Eighty eight percent. I’m going to show you the page it comes from, and then I’m going to show you that same page saying thirty seven percent.

I’m Rory Safir. If any of this is useful, take it and run it. And if you want the manual editions or the studies from this episode, email me at Rory at the Safir lawyer dot com, and I’ll send them to you. No charge and no strings.

Take care of yourself, and take care of your clients.