A 10th person has died of brain cancer after having worked with leather-processing chemicals at Rohm & Haas, bringing the total number of mysterious brain tumors at the Philadelphia chemical-maker to at least 20 in the last four decades.

The family of Olivia Ranalli, who died in March at 60, filed a wrongful-death lawsuit yesterday in the Philadelphia Court of Common Pleas.

Company spokesman Syd Havely said he had not seen the lawsuit but reiterated that a 2004 in-house study found no link between brain tumors and exposure to leather chemicals - or any other chemicals.

He acknowledged that the company has not studied the entire population of employees who ever worked with leather chemicals. The 2004 study included only those who worked at a research facility in Spring House, Montgomery County, where seven of the 10 victims from the leather unit had worked.

Finding an epidemiological link between cancer and exposure to chemicals is notoriously difficult, as the number of cases is rarely high enough to establish statistical significance. In particular, little is known about the causes of brain tumors.

Yet by any measure, the number of victims in the leather division, which makes chemicals to cure and dye leather, seems high.

Nationwide, the rate of brain cancer is about 6 cases per 100,000 people each year. At most, several hundred people worked in the leather unit throughout its history, according to a former executive who asked not to be named. And now 10 of them are dead of brain cancer.

"That's unbelievable, isn't it?" said Norma Iatarola, Ranalli's sister.

Ranalli was a Rohm & Haas saleswoman from 1974 to 1980. She was exposed to chemicals at Spring House and during visits to clients, according to the lawsuit, filed by attorney Aaron Freiwald.

Asked if the company might do a study specifically on leather chemicals, Havely said he did not know. A message left for company epidemiologist Arvind Carpenter yesterday afternoon was not immediately returned.

"Our approach has been, and continues to be, to look at the effect of all of the workplace chemicals used at Spring House," Havely said.

Seventeen of the 20 known brain-tumor patients worked full time at Spring House. Of the 20, 16 developed brain cancer. Four had benign tumors.

Carpenter announced results of his initial Spring House study in January 2004, telling current and former employees he had found no cancer risk associated with chemical exposure at that Montgomery County facility.

A different picture emerges in a document Freiwald obtained during the discovery process of a related lawsuit against the company.

In a draft summary nine days before Carpenter announced his results, Havely wrote that there was a link between brain cancer and two chemicals used by a handful of the tumor patients:

"In the analysis of chemicals data, two chemicals showed statistical significance: N-Nitrosodimethylamine and 1,3 Propane Sultone. Nitrosodimethylamine has appeared as a chemical risk factor in the published literature, but 1,3 Propane Sultone has not. We are examining the use and application of these chemicals among employees."

That paragraph was deleted from the final letter sent to employees. When asked about it, Havely said that the paragraph was incorrect, and that he had written it before seeing Carpenter's full analysis.

"I was trying to get stuff ready in layman's terms, in a layman's letter," Havely said.

In previously undisclosed study data reviewed by The Inquirer, both of those chemicals had "odds ratios" above nine. That is, a person who worked with the chemicals was seemingly nine times more likely to develop brain cancer than someone who did not.

The findings were statistically significant by one measure, called a 90 percent "confidence interval." But they were not significant by another measure, a variable known as the "p" value.That means they were inconclusive, said Timothy Rebbeck, an epidemiology professor at the University of Pennsylvania School of Medicine who was not involved with the study.

"When I see something like that, it's provocative. It means something," Rebbeck said. "But it usually means you need to do more . . . before you can really be convinced that it's real."

A better picture might have emerged with a larger sample size, Rebbeck said.

Yet even if the numbers were inconclusive, Freiwald said, the company still should have disclosed them. In January 2004, Carpenter gave employees and retirees a Power Point presentation that did not include the complete data. More than three years later the results have yet to be published. Company officials say they want to do a follow-up study first.

"Why couldn't they just disclose what they found?" Freiwald asked, questioning whether the paragraph was dropped because officials were "massaging the numbers" in that final week.

Havely said the paragraph was merely his mistake. He wrote the draft after seeing the confidence intervals but took it out after Carpenter showed him the p-values.

Only a handful of the 20 patients were known to have worked with the two chemicals, neither of which is used in leather processing. Some of the leather employees did use vinyl chloride, which has been linked with brain cancer. But the company study did not find any such link.

The leather division dates to the very origin of the company a century ago, when German entrepreneurs Otto Rohm and Otto Haas developed a synthetic enzyme for treating animal hides.

Previously, leather tanners had relied on animal excrement for this purpose. The synthetic version, called Oropon, was an almost immediate success when it was introduced in 1907, in Germany, according to the company's Web site.

In 1909, Otto Haas then traveled to Philadelphia to establish a branch office, which later was established as a separate U.S. company.

Today, it has annual sales of $8 billion.

Contact staff writer Tom Avril at 215-854-2430 or tavril@phillynews.com.