S. Fred Singer's own private reinterpretation of the 1972 Anti-Ballistic Missile Treaty {letters, Feb. 16}, like the other legal contortions underlying the treaty's so-called broad interpretation, ignores the obvious and bungles the fine print. As physicist Wolfgang Pauli once said of a particularly incompetent theory, it's "not even wrong."

Article 2 of the ABM Treaty provides the operational definition of an ABM system as "a system to counter strategic ballistic missiles or their elements in flight trajectory." Although the article states that such systems "currently" consist of ABM launchers, interceptors and radars, it does not limit the treaty to those systems. Article 2 goes on to categorize ABMsystem components as being either"(a) operational; (b) under construction; (c) undergoing testing; (d) undergoing overhaul, repair, or conversion; or(e) mothballed." But it is ludicrous to think that only components in one of these states at the time the treaty was signed are affected.

Mr. Singer's argument is equivalent to saying that laws banning machine guns do not apply to weapons built after the laws were passed, or that special prosecutors can only investigate activities that were under way when the relevant statutes were enacted. (Maybe Mr. Singer should join Oliver North's legal team.) Quite to the contrary -- the ABM Treaty is very clear as to its period of validity. Article 15 says that "this Treaty shall be of unlimited duration" unless either of the parties formally opts out. Mr. Singer's conclusion that the treaty's listing of possible states of ABM development does not apply to exotic SDI technologies because they were then "in the early research stages" is absurd, unless he means technologies so exotic that they can be tested without ever "undergoing testing," built without ever being "under construction" and deployed without ever becoming "operational." GERALD L. EPSTEIN Kensington