Which offers higher HDTV quality: a 2K digital-cinema projector or a 2/3-inch-format box lens? That seemingly unanswerable question is currently being debated at the highest levels of some cultural institutions, but it begs another. Does it matter?

Consider HDTV. It stands for high-definition television. But what does that mean, a screen depicting a dictionary on a ladder?

As with many deep questions, a government committee was formed to deal with it, and, under the heading HIGH DEFINITION TELEVISION, they reported the following: "The degree of definition it is essential to obtain is necessarily a matter of opinion, but the evidence received and our own observations lead us to the conclusion that it should not be less than 240 lines per picture, with a minimum picture frequency of 25 per second."

If that seems a rather low number of lines for HDTV, consider that this particular government committee submitted its report to the British Parliament in January of 1935. Just over four years later, Broadcasting magazine wrote of the 1939 New York World's Fair, "The exposition's opening on April 30 also marked the advent of this country's first regular schedule of high-definition broadcasts." In that case, HDTV meant 441 total scanning lines (in comparison, today's 1080i HDTV has 1,125 total scanning lines).

Clearly, the definition of HDTV has been a moving target over the years. What was called high definition in the 1930s is today called merely television.

There's even a broad range of HDTV definitions today. One currently used professional HDTV camera has just 960 light sensors across the width of one of its imaging chips; another has 5,760.

One of those chips has color filters on it; the other is one of three, each dedicated to a single color. One is much larger than the other. One uses a diagonal spatial offset between colors.

Much could be written about the pros and cons of the different techniques, but they are all intended to provide what is today considered HDTV. Will today's HDTV become tomorrow's ordinary television, only to be replaced by what is today called ultra-definition television? Will that be replaced by hyper-definition television?

When will the cycle of replacement end? Surely, once human beings can't tell the difference between a television image and reality, we will have reached the limit. Won't we?

Consider something theoretically easier to understand than HDTV: sound recordings. Is it possible to record and reproduce sound so accurately that it is impossible for listeners to distinguish recorded from live?

At least one manufacturer claimed that capability. It ran tests all over the country. In small spaces, listeners would be blindfolded while comparing live music to recordings. In major concert halls, a singer would stand next to the sound-reproduction system. In mid-song, the lights would go out. When they came back on, the singer would be gone, leaving only the equipment providing the sound.

A major daily newspaper sent a reporter to one such test. "It did not seem difficult to determine in the dark when the singer sang and when she did not," he wrote. "The writer himself was pretty sure about it until the lights were turned on again and it was discovered that [the singer] was not on the stage at all and that the new Edison alone had been heard."

The new Edison? That review appeared in 1919, and the Pittsburgh Post reporter had been listening not to an oversampled, high-bit-depth digital recording but to an Edison phonograph. He was not alone in being unable to tell them apart.

That doesn't mean the differences were beyond human hearing. In those early days of sound reproduction, it was so exciting to hear anything at all emerging from a machine that audiences probably didn't listen too critically (and one "tone test" singer admitted in 1972 to having trained herself to sound like a phonograph recording). As sound reproduction has improved, however, so, too, has our ability to distinguish it from reality.

Something similar occurred with multichannel sound. Stereo dates back at least to 1881, and, even then, listeners knew it offered spatial localization. But, when stereo was commercialized more than half a century later, it was supposedly so perfect at reproducing acoustic environments that it was indistinguishable from reality.

There was a seeming physiological basis to that claim. Humans have two ears; stereo has two speakers. Q.E.D.

Stereo speakers, however, are not pressed against ears. In a normal stereo listening environment, both ears hear both speakers, and moving slightly can shift the apparent acoustic position of a sound.

Today's 5.1-channel surround sound fills in many holes, but it doesn't really help vertical localization. The ultra-definition television (UDTV) system shown by NHK (the Japan Broadcasting Corporation) on the exhibition floor of the National Association of Broadcasters (NAB) conventions has 22.2 channels of surround sound, but no one has yet insisted that number makes its acoustic reproduction indistinguishable from reality.

What about the picture definition of UDTV? It has 16 times more picture elements than does the highest-resolution version of current HDTV. Is that enough?

Consider the Snellen chart. It's the familiar eye-test image with the big letter "E" on top.

Actually, you're not supposed to think of that "E" as a letter. It's an "optotype," a character specially designed for the measurement of vision. Each line of the "E" or space between the lines has exactly the same thickness. When a Snellen chart is viewed at the correct distance, an "E" on the line for normal (20/20) vision will cover a visual angle of five arcminutes (five 60ths of a degree).

As an "E" has three horizontal lines and two spaces between them, that means that each line or space on the 20/20-vision line covers just one arcminute, a 60th of a degree of arc. That has been taken to mean that human visual acuity tops out at one arcminute.

Now consider a television picture. It consists of scanning lines. If you can see the scanning lines, they are interfering with your ability to see the picture, a case of not being able to see the forest for the trees.

If human visual acuity tops out at one arcminute, then, for the active (picture-carrying) scanning lines in ordinary U.S. television to be just barely imperceptible, they should each cover an angle of about one arcminute. If bigger, they would be clearly visible; if smaller, viewers wouldn't get maximum resolution.

Not counting overscan (common magnification of the image beyond the edges of the screen), there are about 480 active scanning lines in a traditional (non-HDTV) U.S.-standard TV picture. For each to be one arcminute, all 480 would be eight degrees in total. The picture of a 25-inch TV set in the traditional 4:3 shape is 15 inches tall. Viewed from the Lechner Distance (nine feet, the distance that a researcher at RCA Laboratories once found was the typical home viewing distance in the U.S.), that picture would subtend an angle just under eight degrees.

It's almost as though U.S. television was designed with that visual-acuity criterion in mind. Unfortunately, there are two problems with that theory.

First, the 25-inch TV set is a relatively recent phenomenon. Decades after the introduction of commercial television in America, an advertising campaign for radio that indicated it was a medium that stretched the imagination was able to answer a question about whether television stretched the imagination with the punch line, "Up to 21 inches, yes." According to the Consumer Electronics Association, as recently as the year 2000, the average size of a TV screen sold to a U.S. dealer was smaller than 25 inches.

More significantly, one arcminute is by no means the limit of human visual acuity. Consider Sirius, the brightest star in the sky.

If you have ever seen any star, your visual acuity is sufficient to see Sirius. But its angular diameter is roughly six milliarcseconds. That's six thousandths of one 3,600th of one degree of a circle's arc or about ten thousand times finer than the 20/20 vision measured on a Snellen chart.

If Sirius were a single pixel on a 25-inch 4:3 TV set viewed from the typical nine-foot Lechner Distance, that TV would have close to five million scanning lines (UDTV has just 4,320). And, when you look into the night sky, you can probably see a lot more stars than just Sirius. So, how can the visibility of Sirius coexist with the use of a Snellen chart?

The answer is contrast. Sirius is a star 25 times more luminous than our sun, and we view it against the blackness of interstellar space. That's a lot of contrast.

A television viewing environment is something else entirely. With it turned off, look at your television set and describe the screen. Depending on the model, you might say it's brown, green, or gray, but chances are very good that you wouldn't call it black. It's not black because it's reflecting the light in the room.

It will continue to reflect that light when the TV is turned on. When a manufacturer lists a contrast ratio for a display, that figure supposes there is no room light whatsoever. If a video display can deliver even a whopping 1,200 candelas per square meter (about 350 footlamberts) at peak white, in a typical living room environment that might mean a maximum contrast ratio not of 10,000:1 or even 100:1 but perhaps just 25:1; in a bright living room, it could drop to perhaps 5:1.

Under low-contrast conditions, human beings cannot perceive detail as fine as the star Sirius. In fact, we can't even make out scanning lines subtending an angle of one arcminute. Detail might have to be between 12 and 30 arcminutes to be seen in poor contrast.

That's why, even though thousand-line television was proposed back in the 1930s, few saw a need for it. With 25-inch (or smaller) TV screens viewed from a distance of nine feet in low-contrast conditions, even ordinary television provided more resolution than could be perceived.

Today, of course, screens are much larger. At the Consumer Electronics Show this year, Panasonic showed a 150-inch (12.5-foot) plasma TV with a resolution of 4096x2160. It is clearly not meant to be viewed in a typical living room, which means lights might be very dim where it is viewed and, therefore, contrast high. Under high-contrast conditions, humans are capable of perceiving Sirius.

After about three-quarters of a century, TV also finally seems to be catching on in movie theaters. The Metropolitan Opera's Live in HD cinemacast of Roméo et Julliette last December achieved the equivalent of 11th-highest U.S. theatrical box-office gross for the weekend, despite only a single showing. The ongoing nature of the series has led more theaters to carry it and more organizations to try their hands at cinemacasting HDTV.

That led to this month's first question. One organization says it offers more quality because it delivers only to theaters with "2K" (slightly-beyond-HDTV resolution) digital-cinema projectors, but it has used 1/2-inch-format imagers and "barrel" lenses. Another organization accepts other HDTV projectors but uses only 2/3-inch-format imagers, usually with higher-quality "box" lenses.

Which offers higher quality? That question might still be unanswerable. It's something like asking whether proper tire inflation or proper spark timing offers better automotive performance. But quality does matter. On a giant screen in a dark theater, it's nearer getting Sirius.