Just as auto buffs love to argue the relative merits of Mercedes and BMW, just as skiers will debate for hours over the burning question of Aspen versus Vail, so computer jocks everywhere love to sink their teeth into the endless argument over which PC is better: IBM or Macintosh.

The MacFanatics point out, accurately, that IBM has "borrowed" a lot of ideas that surfaced first on the Macintosh: the mouse, the "desk top" interface and "point-and-shoot" file selection. Big Blue believers respond, accurately, that the Mac has "borrowed" lots of things from the IBM PC as well: function keys, high-resolution color display, modular design and expansion slots.

As long as the debate stays at that level, the question will never be resolved. But what if somebody set out, in a careful, professional way, to quantify the differences? Somebody did. The National Software Testing Laboratories (800-223-7093), an established independent evaluation firm that publishes a respected monthly publication called Software Digest, recently set up a series of laboratory tests to compare price and performance ratings for various models of Apple's Macintosh and IBM's PS2 lines.

NSTL paired similarly priced computers against each other: the Mac SE against IBM's Model 50Z, the Mac IIci against the IBM Model 70-A21, etc. Although we have some quibbles with the types of tests used, we can't quarrel with the rigor or objectivity of NSTL's procedures. Still, the results were disappointing. Nobody won.

Having tested all kinds of IBMs and Macs on various kinds of software, the laboratory reached a conclusion that falls into the kissing-your-sister category: The great competition came out in a tie.

"It's very difficult to determine an overall winner," NSTL concluded. "Striking similarities and few disparities exist between Macintosh and IBM systems."

In a sense, this makes sense. As anyone can see, the traditional differences between the basic Macintosh and IBM (DOS) architectures are disappearing. The two systems are borrowing effective ideas from each other all the time. Already, there are many programs -- such as Pagemaker and WordPerfect -- with virtually identical interface and command structures on the two machines. AutoCad, for example, ships the same reference manual for both its Mac and DOS versions.

So it wasn't surprising that the lab tests found nearly identical speed and performance numbers using Microsoft's Excel spreadsheet on the Mac IIci and the IBM PS2 Model 70. Similarly, on the FoxBase data base program, comparable models of Mac and IBM turned out to be roughly equal in speed and processing power.

One problem with this comparison is that the NSTL mainly tested for speed. How long did it take each computer to draw a chart in Excel, or spell-check a long document in Microsoft Word? These are legitimate questions (and questions that can easily be tested in a laboratory setting). But they miss the point of why most people pick a computer.

One reason people buy the Macintosh is that it is simpler and more fun to use than any other PC. The laboratory tests didn't go into the question of "how it feels," but that has to be a real consideration for any buyer.

On the other hand, a key reason people choose IBM computers, or other DOS machines, is that there's a much greater variety of software and peripherals in the DOS world than in the Macintosh sphere. And DOS prices tend to be lower. If you need a hard disk, say, or a full-page monitor, you'll have to pay more for the Macintosh version. But this salient point was ignored in the head-to-head performance tests.

You might say that the real winner of this Mac vs. IBM face-off was the competitor who wasn't there -- that is, the DOS clones. NSTL chose only to test products made by Apple and by IBM. Actual computer buyers have a vastly wider choice: Macintosh, IBM, or any of the thousands of "clone" machines that run DOS or its big sister, OS2.

While NSTL found roughly equal pricing between comparable models of IBM and Macintosh PCs, the DOS clones are almost always cheaper than either IBM or Mac. If you replace the IBM Model 70-A21 in the tests with a clone machine from, say, Dell or Gateway or AST, it would perform just as fast as the Mac IIci -- but at about half the cost.

Judged strictly on price and performance benchmarks, then, we'd conclude that DOS clone machines offer the most bang per buck in the personal computer world. But to look strictly at price and speed is to ignore the Macintosh's ease-of-use and IBM's hard-earned reputation for quality and service.

So despite the head-to-head laboratory tests, we can still argue 'til the cows come home over which PC is the best. And we surely will.