EVEN THE BEST TRIAL LAWYER WILL tell you that juries are like snakes: They’re hard to get ahold of and not necessarily fun when you do. That’s why law schools concentrate on teaching the law. Aside from picking up a few rules of thumb—Episcopal clergy side with plaintiffs; Episcopal churchgoers side with defendants—law students have to learn about juries the hard way: by trying lawsuits after they graduate. Even so, juries remain enough of a mystery to give rise to an entire industry of consultants available to predict the best argument for your case, provided the client is willing to pay for it. When I litigated for an Austin trial firm in the mid-nineties, I helped pick two juries; one we misread and one we nailed, to the tunes of $27.5 million and $3.5 million, respectively. I liked to tell people that when I left the firm I’d gotten up from the table with a record of one win and one loss, as opposed to being down $24 million. It was never completely clear to me, however, why one jury had gone so well and the other so wrong.

I did not expect to get any more insight once I retired and became a recovering lawyer. The only way I would learn more about juries would be by serving on one, and that wasn’t likely to happen. Lawyers don’t get picked for juries. They think they know too much about what’s going on at the trial, and they tend to interpret the law—a job that belongs solely to the judge—as well as the facts. So when I reported for jury duty in county court in January, I felt sure I wouldn’t be chosen. I was even more certain when our eighteen-person panel was led into the courtroom; I was candidate number twelve for a jury of six. Tradition and probability indicated that I’d be back at the office just after lunch.

So much for probability—I was the last juror picked. In the meantime, I had listened to the prosecutor move slowly through voir dire, the pretrial procedure in which the attorneys for each side quiz panelists to determine what kind of jurors they would make. The defendant was accused of driving under the influence. “If a law officer were to take the stand,” began the prosecutor, “and he were to tell you that on a roadside stop he had seen all the telltale signs of drunkenness in the person pulled over—the slurred speech, the stinking breath . . .”

“Objection, your Honor. He can’t say that.”

“Overruled.”

“. . . the poor balance, all of it—would you think that that officer’s testimony was more credible than someone else’s simply because you knew that he, the law officer, had years of experience and was specifically trained to evaluate these very signs?”

In other words, I thought, he doesn’t have much scientific evidence: no Breathalyzer results, just cops who thought the defendant was drunk versus non-cops who thought he was not. The prosecutor needed us to believe the police officers. But one of the panelists, a blond estate planner who looked fresh out of law school, saw something more after looking at the men at the defendant’s table.

“I don’t think you can automatically make those assumptions,” she corrected the prosecutor. “Statistics show that African Americans are pulled over in numbers far greater than whites, and in a case like this, in Texas, with an African American defendant, I’d want to be careful that there wasn’t something more going on when the police made the stop.”

Nice try, but as the judge pointed out, the African American man at the defendant’s table was one of his lawyers. The defendant was white. There was a collective wince in the courtroom, but the judge and the attorneys, who have to woo reluctant jurors for every trial, just smiled and moved on. The estate planner smiled too when she wasn’t named to the jury. But I was.

We were a typically random group—four women and two men—that, in my opinion, was slightly inclined to favor the defense. A woman who marketed promotional doodads had complained during voir dire about an unfair DWI conviction in her past. Another woman coordinated projects for a large construction company and indicated, like most of the rest of the panel, that she would never blow into a Breathalyzer on principle alone. A cheerful homemaker stated that a primary effect of alcohol was “It makes people more affectionate.” I noted that she was pregnant.

Our law-and-order bloc was a nurse with the Texas Youth Commission, who said she couldn’t drive after even one drink, and a “forensic architect” with a black-and-white worldview, who made his living testifying as an expert witness in construction litigation. I considered myself a neutral faction of one.

The prosecutor opened his case the next morning. His problem rapidly became apparent: After the defendant had allowed the police officers to gauge his eye movement, he had refused all other field sobriety tests. He was a 55-year-old man who could have easily passed for 75, and he had told the officers that his feet and back bothered him too much to walk heel-to-toe in a straight line or balance on one leg. He had also declined to blow into a Breathalyzer, so the only evidence left to the prosecution was the testimony of the officers involved and a videotape of his roadside refusals and drive to jail, and the video wasn’t even complete. The judge had made the prosecutor mute one 25-second segment of dialogue.

The police officers testified exactly as the prosecutor had promised, and the defense countered with a strategy of pure aggravation. The defendant’s lead attorney made repeated objections and motions for a mistrial that the judge quickly overruled, and he labored to catch the police officers in trivial inconsistencies. His examination of the second officer was spent rewinding the video to sections where the officer sounded, according to the defense, “sarcastic and smart-alecky.” When he did make a good point, getting the officer to admit that there were less-demanding tests he chose not to give—such as asking the defendant to say his ABC’s—I wondered if the rest of the attorney’s performance was intentionally inept to highlight this small victory.

He called some likable witnesses, two pleasant ladies who said they had been at a birthday party with the defendant before he was pulled over and supported his claim that he’d had only three beers. He hadn’t seemed drunk to them. The defense also produced some medical records that appeared to back up the alleged feet and back trouble. The defendant did not take the stand.

On a good day the trial would have taken all of two hours, but between the defense lawyer’s objections and the judge’s frequently shooing us from the courtroom to tend to other business, it was eight hours from the ten-thirty start of the trial until the attorneys rested their cases. We spent half of that time holed up in the jury room, careful not to talk about the case. By the time the attorneys made closing arguments, we had run out of questions about each other’s job and mutual friends, and we were ready for dinner and a chance to go home. We took a quick straw poll to see if we were close to the unanimous vote we needed, and the results were near my initial estimate: Four “not guilty” and two “guilty,” with the disagreement coming exactly over the prosecutor’s initial hypothetical question—Was the testimony of police officers alone enough to rely on? For the construction coordinator and the forensic architect, the defendant’s shaky refusal to walk the line, as captured on video, was enough. The “not guilty” bloc, which included me, wanted more. The arresting officer’s first comment to the defendant—”What’s that on your jacket there, a little vomit?”—indicated that he had made up his mind that the driver was drunk before he’d talked to him. With the driver’s witnesses saying he was sober, all we had to go on was our impression of the video, and all we saw there was an old man claiming a bad back and feet. We debated for an hour, watched the video again, and then dug in our heels.

“Just because the cop expected him to be drunk,” argued the architect, “doesn’t make him sober.”

“That’s not the point,” I said. “It doesn’t mean that he’s sober; it just means the cop’s opinion isn’t enough. He had that opinion before he even met the guy.”

“But the cop would have changed his mind if he hadn’t been right. Suppose he had gone there looking for a green car and found a white car instead. He wouldn’t still insist it was green.”

“That’s got nothing to do with anything. Green cars are green cars, but ‘Is he drunk?’ is an opinion.”

It was close to nine-fifteen. The judge’s earlier talk about having pizza delivered had been replaced by directions to the vending machines, none of which would take dollar bills. The juror with the DWI conviction, whose husband had instructed her not to get picked for the jury in the first place, said she would never vote to convict. Shortly after ten o’clock, the judge declared a mistrial.

On the way out of the courthouse, I rode down the elevator with the prosecutor. I wanted to ask him about the part of the videotape the judge hadn’t allowed us to hear. I knew the judge would have cut out anything overly prejudicial, which in this case would have been prior trouble with alcohol. But she had told us not to speculate on facts we weren’t given, and I’d put it out of my mind. I was hoping now that I hadn’t overcompensated, putting a habitual drunk driver back on the street. I asked the prosecutor what had been deleted.

“The officer asked him if he’d ever been arrested for DWI before, and he said this was his third arrest.”

“His third arrest?”

“Yeah,” he shrugged. “Once these guys figure out that they don’t have to take the Breathalyzer test, we don’t have a chance. Without that prior-arrest history, we never get any of these guys.”

I watched him walk out. He’d done everything he could to get us to see a drunk man in the video. He’d asked us to think about what a drunk looked like and who best could tell. But until that conversation, I just hadn’t seen it.