fter a humiliating deflation this February, human dignity is on the recovery path, at least as measured by the number of genes in the human genome.

Two new estimates put the likely number of human genes at around 40,000, up by a third from the estimate of about 30,000 in February by the two teams of scientists who decoded the human genome. The low estimate still has its defenders.

The 30,000 number suggested that evolution had not found the design, operation and maintenance of a person much more complicated than the job of running the microscopic round worm (19,000 genes), the fruit fly (13,000 genes) or the mustard cress plant (25,000 genes), the three other multicellular organisms whose genomes have been analyzed.

Until the human genome was decoded, the textbook figure for the number of human genes had been 100,000, putting a safe distance between people and the little creatures on which the genome decoding methods were first tested.

But 100,000 was just a guess. When the 3.2 billion letters of the human genome were decoded, computer programs set to analyze the sequence found that 1 percent of the DNA consisted of genes. Both the government-financed consortium of academic centers and its rival, Celera Genomics of Rockville, Md., concluded there were 30,000 or so well- documented genes, though both allowed that the number might grow.

Dr. Michael P. Cooke, Dr. John B. Hogenesch and colleagues at the Genomics Institute of the Novartis Research Foundation in San Diego have now compared the two sets of human genes from Celera and the consortium, finding that 15,000 genes are common to both. The fact that the two sets do not entirely overlap has not been generally known.

Based on a test of their own, the Novartis biologists believe that most of the genes found by Celera alone, and by the consortium alone, are real. If all the human genes found by only one of these groups are added to the 15,000 found by both, the total comes to around 40,000.

Dr. Cooke said his next project was to test whether all 40,000 of these predicted human genes were real.

Separately, Dr. Christopher Burge of the Massachusetts Institute of Technology said he had an article in the journal Genomics suggesting that the total number of human genes was around 40,000. Dr. Burge is the author of Genscan, one of the principal programs for predicting which regions of the genome contain genes.

Dr. Cooke and Dr. Hogenesch said they were less interested in the precise number of human genes than in making sure they had the full set, however many it might be, since researchers trying to understand some aspect of the body's function could be led astray if some vital gene was missing from the catalog.

The importance of cataloging every human gene is to understand how human cells work in health and disease and to devise gene-based remedies for many ailments.

The number of genes dictates the size of that task, since it would presumably take longer to understand the roles of 40,000 than of 30,000. Obtaining the precise number is also a test of how well biologists understand the human genome.

Now that the number may be ascertainable, biologists are being more cautious in their predictions.

"People were throwing numbers around, and now there is complete silence on the subject," said Dr. Burge, suggesting there was a fear of looking foolish with the wrong estimate.

In May last year, with estimates of the number of human genes ranging all over the lot, scientists meeting at the Cold Spring Harbor Laboratory on Long Island organized a sweepstakes for bets on the exact tally, with the winner to be chosen in 2003.

It is now not so clear that the real number of human genes will be known any time soon. With all the estimates out there, "it has to be becoming clear to people it must not be simple," said Dr. J. Craig Venter, the president of Celera Genomics.

Dr. Venter said that the Novartis scientists were the first to point out publicly that Celera's predicted set of human genes and that of the public consortium did not overlap much, a fact that means the consortium "has missed a lot."

Dr. Eric Lander, chief author of the consortium's report, was traveling and unavailable for comment.

Dr. Venter said he disagreed with the Novartis scientists' method and did not plan to revise his tally of 26,000 well-documented human genes. He also found other more doubtful ones, and with further research his number could go up or down by 5,000 genes.

"It will probably take 5 or 10 years to have a really accurate count plus or minus 100 genes," he said.