When Secretary of Defense Caspar Weinberger promised earlier this month to arrange an intelligence briefing for Pentagon reporters, it represented one of those rare occasions when the government and press try to make ends meet. The secretary had accepted what reporters have called a "non-adversarial" challenge for a show-and-tell of the evidence behind administration admonitions that America is vulnerable to improved Soviet military readiness, in order to evaluate, as one put it, the "we've-got-to- rearm drumbeat" of President Reagan and others. Some evidence suggests that, in order to deliver, Mr. Weinberger had to persuade a reluctant intelligence community.

Between then and last week, as you may have read, official zeal took over and almost let nonsense run away with common sense, violating an old law-- not of the iron variety that holds that a little constructive ambiguity never hurt anybody but that government/press relations flaps can be born, but never made.

As a condition for attending the briefing, reporters were asked to sign what appeared as a legalized ("I take this obligation without mental reservation . . .") oath of secrecy. It would hold them to "not divulge in verbal, written, broadcast or other discourse" what they learned -- this was explained orally to preclude informing their editors -- and prohibited any reference to "the generic source of information." Then, they were requested to report any approaches for any part of it straightaway to the Defense Department. Finally, if accepted, the document was to be countersigned by a witness.

It may be unnecessary to say that reporters would not be caught in the same room with these terms. (Post reporter George Wilson was on another assignment at the time. Learning of the conditions beforehand, The New York Times refused to send its correspondent). Those present, representing more than a dozen major new organizations and not forewarned unanimously refused to sign the paper, and did so a second time after officials amended it. In the midst of this, according to a reporter, one official remarked, "Why don't we call this an abort." That suggestion, everyone now agrees, should have been acted on while writing the paper.

Discussion with various officials and most of the reporters suggests something Aesopian to the encounter. While justifying the document and the need to protect sources and methods, the ranking military officer was described as telling reporters, "For this day, all would have security clearances," overlooking that that's not what news people are in business for. Public affairs officers, who have endeavored to provide more material for the press, insist that the written conditions were a "sine qua non" from the intelligence community.

Finally, and sensibly, the document was withdrawn, but only after officials confronted what might have been anticipated: a self-inflicted public relations wound.

The briefing was conducted then under the more conventional off-the- record rule, meaning no immediate stories, no identification of briefers and, above all, no reference to sources and methods of intelligence gathering. Practically, this does permit future use of the data learned as background in related news material. The connoisseurs recognize this as "deep background."

Contrary to some news reporting and editorial comment, reporters did not give oral acceptance to what they wouldn't sign. While not an everyday affair, the eventual format and terms were familiar to journalists who cover national security affairs. Reflecting reportorial opinion generally, Walter Mossberg of the Wall Street Journal said, "What I agreed to was a standard press ground rule under which I'm at liberty to use material on Soviet military capability."

Nor is it correct that the administration suckered reporters to a briefing in order to declassify intelligence information for political purposes. It was the press' initiative. If in the end it serves both sides, that's not bad.

What's bad are press secrecy oaths. Institutional memory may be as flawed in Washington as anywhere. But there's enough around to certify what was tried was a first, and like all bad precedents, should not be tried again.