The gas used to subdue terrorists and their hostages in a Moscow theater two weeks ago most likely was an aerosol form of carfentanil, an extremely potent narcotic normally used to used to sedate big game animals.

The gas, which appears to be responsible for the deaths of 123 of the 128 hostages killed during the crisis, may have also contained halothane, an inhalational anesthetic that has been used in surgery for nearly 50 years. While each compound can be lethal on its own, mixing them is likely to greatly increase the hazard.

Those are among the conclusions reached by numerous scientific experts based on eyewitness reports, observations of physicians who treated the stricken hostages, the condition of the patients, statements by Russian officials and several laboratory tests performed on survivors.

Russian special forces troops stormed the theater on Oct. 26 after pumping gas through ventilation ducts of an auditorium where several dozen Chechen militants were holding more than 800 people who were there to see a play. Nearly all the militants were killed, some by the gas and others who were shot after being subdued by it.

The Russian health minister, Yuri Shevchenko, said the gas contained "derivatives" of fentanyl, a compound that acts like morphine but is about 80 times more potent. He would not name the constituents.

Tests conducted on two surviving German hostages -- one so severely gassed that she required breathing assistance with a ventilator for several hours -- found no evidence of fentanyl. Carfentanil is a fentanyl derivative sold under the trade name Wildnil. It would not be detected in a test for fentanyl.

Although not approved for human use, carfentanil has two attributes that make it a likely candidate. It's about 8,000 times as powerful as morphine -- so potent that it might be feasible to deliver a disabling dose to hundreds of people in a large architectural space. Carfentanil also has an unusually wide safety margin, which is why it's the immobilizing agent of choice when delivery is by the unpredictable route of a rifle-fired hypodermic dart.

"It was the safest drug that you could have used in this fashion and hope to have high percentage of survivors," said Theodore H. Stanley, a professor of anesthesiology at the University of Utah Medical Center and an authority on opioids, compounds that function like morphine.

Stanley's research with animals helped lead to carfentanil's licensing a decade ago as a veterinary sedative. His laboratory has also been part of the government-sponsored effort to find compounds -- including even more potent opioids -- that could potentially be used for crowd control.

"Of the things that are generally known to pharmacologists, carfentanil would be the best guess" as the substance used in the Moscow crisis, said James H. Woods, professor of pharmacology at the University of Michigan and also an opioid expert.

"If one wanted to look at a combination of the most potent material as well as availability . . . carfentanil would be a good candidate," said Parker Ferguson, a former government chemist who is now an independent consultant on chemical and biological terrorism.

Independent of the health minister's testimony, there are reasons to believe that an opioid drug was a main constituent of the gas.

Shevchenko said that "1,000 antidotes" (which he would not name) were on hand at the time of the rescue. Fentanyl and its relatives are rapidly reversed by naloxone or naltrexone, two so-called opioid antagonists given routinely in emergency rooms for heroin overdose.

Two Moscow physicians who spoke on condition of anonymity said that many patients had classic signs of opioid intoxication: pinpoint pupils, unconsciousness, depressed breathing. (Some also had bluish skin, probably a consequence of reduced amounts of oxygen in their blood because of very slow respiration rates induced by the gas.) Some patients also vomited, which is common, although not universal, in opioid overdose.

All hostages treated at their hospital were given naloxone, and none failed to respond, one of these doctors said.

At least one other physician, however, said many cases were perplexing.

Yuri Goldfarb, chief toxicologist at the Sklifosovsky Emergency Aid Research Institute, said 42 people from the theater were treated there and many were conscious and had no difficulty breathing. A few had signs of mild liver damage, although he said he couldn't necessarily attribute that to the gas. Despite the institution's expertise, no toxicological tests were done.

"To conduct such tests we have to know approximately what we're looking for, and we didn't know what to look for," Goldfarb said. "Besides, we didn't have the technical means to conduct such tests."

He added, however, that "in general, I can conclude a number of symptoms were different than a classic opiate overdose."

Both German survivors -- a woman about 18 years old and a man in his fifties -- had traces of halothane in their blood when they were tested in Munich about 36 hours after the rescue. The woman, the more severely affected of the two, also had halothane in her urine, indicating she had received a higher dose.

Halothane can be absorbed and re-released by rubber and some plastics. Rare instances have been reported in which a person has picked up trace amounts of the gas after being put on a ventilator previously used in surgery on a patient anesthetized with halothane. This might explain the woman's test results. The German man was never on a ventilator, ruling that out as a possible route of exposure for him. "I don't know that they used it. All I know is that I found it," said Ludwig von Meyer, professor of forensic toxicology at the Institute of Legal Medicine at the University of Munich, who oversaw the tests.

The chief anesthesiologist of the clinic of the Technical University of Munich, where the patients were examined, was more certain. "It can be excluded that they were exposed to halothane in the treatment process [in a Russian hospital]," said Eberhard Kochs.

There are no commercially available tests for detecting carfentanil. Von Meyer said that his laboratory is considering asking Paul A. J. Janssen, a renowned Belgian physician and researcher who first synthesized fentanyl and its derivatives, for advice on how to look for carfentanil in the small remaining samples of blood and urine.

The presence of halothane in the Russian gas mixture could explain why some unconscious patients weren't revived with naloxone, a drug that reverses the effects only of opioids, not other anesthetics. (Another reason is that many patients may have suffered irreversible brain damage by the time they were given naloxone.)

Halothane, however, has a narrow safety margin, making it an exceptionally dangerous tool for crowd control. It's one advantage is that it's a true gas, vaporizing and dispersing naturally at room temperature.

In contrast, fentanyl and its sister compounds are solids. To be distributed through the air, they must be sprayed -- either as fine powders or, more likely, as a mist after being dissolved in water. Such a delivery system is fraught with difficulty.

"It takes a staggering amount of fentanyl to subdue a patient," said David R. Drover, an anesthesiologist at Stanford University Medical Center and a specialist on the clinical effects of anesthesia. "To actually fill a room, even a small room, would have taken a swimming pool-full. No way."

Some hostages reported seeing a thin vapor before losing consciousness. Such reports would be more consistent with the use of a super-potent derivative.

"The fact that carfentanil is 100 times more potent than fentanyl makes it 100 times easier to use than fentanyl -- and that makes it a logical choice," Utah's Stanley said.

A drug's safety is measured by its "therapeutic index" which is the ratio of a lethal dose to an effective dose. The therapeutic index of carfentanil in rats is 10,000, meaning that a lethal dose is about 10,000 times more than the dose required to relieve pain. Halothane, in contrast, has a therapeutic index of about 3, meaning that a lethal dose is only 3 times that required to put someone to sleep.

Another fentanyl derivative -- sufentanil -- has a therapeutic index of 25,000, making it safer than carfentanil. It might have been used. It's only one-half to one-quarter as potent as carfentanil, so a greater volume would had to have been used.

Carfentanil's safety margin may be what Russian officials were referring to when they expressed amazement at the fatalities. Shevchenko, the health minister, said that "on their own, these substances cannot lead to a fatal outcome." A spokesman for the Federal Security Service, Alexander Zdanovich, said there was "a consensus among the experts we consulted that this drug could not have caused death."

However, there is extreme variability between species in response to opioids, especially their toxic effects. Dogs, for example, can withstand doses of some drugs that are immediately fatal in human beings.

A State Department spokesman would not comment on what efforts, if any, are underway to identify the gas. Clothing from several British hostages is being tested at Porton Down, England's military laboratory for chemical and biological defense. A British Embassy spokesman here said results are not yet available.

Baker reported from Moscow.