Charge: driving while intoxicated. At trial, the arresting officer testifies that she stopped D for speeding. While writing out a speeding citation she observed a beer can on the seat next to D. The prosecution offers the beer can as evidence.

Should the beer can be admitted? Does it make any difference whether (1) the beer can is half empty, (2) the beer can is open and completely empty, or (3) the beer can is full and unopened?

For each situation, (1) the can half empty, (2) the can completely empty, and (3) the can unopened, we can imagine different stories that are consistent with guilt or innocence. The empty can: D had just polished it off, or it had been in D's car for a week. The unopened can: D had not been drinking at all, or D had already drunk five cans of a six-pack, throwing the empties out the window, and D had only one left. The half-empty can seems different: We feel pretty sure from its presence that D was drinking at a time proximate to the accident, but how much? If only half a can, it is unlikely that D was drunk, but if D were chugging beers down, as people who drink beer while driving a car might be thought to do, then the half-empty can may show that D was drinking too much. It at least shows that D is the kind of person who drinks beer in a car while driving.

With regard to the empty or the unopened can, does the presence of the can make a material issue more or less probable? Probable in what sense? Evidently, more probable in the minds of the factfinders. But what factfinders? There may be six or twelve of them, and they may have different views about the way the world works. It seems as if "probable'' means that if somehow we could run time backwards and count all past events in which people had an unopened or an empty can in their car, and then could somehow know the percentage of these occasions on which the driver was drinking to excess, then we could say whether the presence of either the full or the empty can made the issue of drinking to excess more or less probable in this case. But of course we cannot know these things. We have only the roughest ideas about such things, and each of us may have a different idea based on our own different life experience.

But because we do not know which of two possibilities, one consistent with innocence, one consistent with guilt, is the more "probable,'' does that mean that we should assume equal possibility and therefore irrelevance? We have no basis for assuming equal probability except our ignorance. In other words, if we could somehow make the fanciful count described above and arrive at rigorous frequency data about the mutual occurrence of drunkenness and the presence of an unopened beer can in the car, it would be exceedingly unlikely that the presence of the beer can would indicate an equal probability of drunkenness and sobriety, or fail at all to correlate with drunkenness as compared with drivers who had no beer cans in the car.

Whether the beer can makes a material issue more or less probable is largely indeterminate, since it depends on the individual perception of the world of the person making the judgment. If reasonable people can and do differ on the question of whether offered evidence makes the existence of the fact sought to be proved more or less likely than without it, what should the judge do? Who provides the relevant measure? The judge or a "reasonable jury''? What should the trial judge or a reviewing court take into account or bring to bear in making relevancy judgments? Logic? Experience? Local idiosyncrasies?

The beer cans problem can be used to develop these questions further. Whether the beer can makes a material issue more or less probable is largely indeterminate, since it depends on the individual perception of the world of the person making the judgment. What then is the relevance judgment that the judge must make at the point of deciding the admissibility of the evidence?

We believe that the correct question is: Would you want to look at the item of evidence if you embarked on the task of trying to figure out what happened? This does not mean that the item must be conclusive, or even that its significance must be obvious. Its significance may very well depend on what other items of evidence you are able to assemble. Yet the nature of the relevance judgment that a judge must make at the point of admissibility is that the judgment must be made without knowing the full context that will be provided by all other items of evidence (subject to the qualification that a judge has some means to "peek" ahead by asking an attorney what other proof he has). This is true for every piece of evidence. So the relevance judgment depends on a rough judgment about how an item of evidence might relate to a material issue, depending on what other items of evidence might be available to provide a context for considering the first,

A person hired to investigate the beer-can case who went to the scene to collect evidence, and who had to sort through all the items in and around the car, saving the things thought to be "relevant" to the investigation and ignoring the things thought to be "irrelevant," would of course have saved the beer can, whether it was empty or unopened. That, we believe, is the FRE 401 concept of relevance. This concept is not to be confused with the overlay of considerations of prejudice introduced by FRE 403. Nor should this concept of relevance be confused with the question of the sufficiency of the evidence to sustain a verdict (directed verdict question). In neither the burned-butt problem nor the beer-cans problem does it follow from the admissibility of the evidence that, assuming no further evidence, a verdict for the proponent of the evidence could or should be upheld.