When the killing of George Floyd drew tens of thousands of protesters into the streets of America’s largest cities, many of those streets had been empty for weeks.

Restrictions had left family members unable to attend the funerals of relatives for fear of spreading the novel coronavirus. Shutdowns sent businesses into bankruptcy.

But after Floyd’s death, the streets filled with people shouting and yelling in proximity — sparking concerns among public health experts and local officials who had been urging people during the pandemic to stay at home or to engage in social distancing.

Now, some public health officials and disease trackers say there appears to be scant evidence the protests sparked widespread outbreaks. Others say that because many states reopened about the same time as the protests, and because of the limits of contact tracing, they simply can’t say for sure.

“I’m about to do a podcast laying out all I don’t know,” Michael T. Osterholm, director of the Center for Infectious Disease Research and Policy at the University of Minnesota, said last week. “And it’s a hell of a lot more than I know.”

As protests were building across the nation, hundreds of public health experts signed a letter arguing systemic racism is a public health crisis, too, and that protests were therefore worth the risk — even as many of them warned that protests could spread the virus. Cities hurried to set up testing facilities near protest sites to identify cases early.

The number of positive coronavirus tests in recent weeks have grown almost unchecked in many parts of the country. Hospitals in Arizona, California and Texas are stretched to the breaking point. Governors are resorting to the once politically unthinkable measure of shutting businesses again. But most experts say the protests are probably not to blame, or almost certainly not the only thing to blame in places where cases are soaring.

Absent a few positive tests among protesters announced here and there, the only major outbreak tied to protests happened in South Carolina, where organizers postponed demonstrations or moved them online after at least 13 people who took part in previous protests tested positive.

Organizer Lawrence Nathaniel posted a video to Facebook saying those testing positive marched in Columbia, S.C., between May 30 and June 17, including six protesters. four organizers of the demonstration and three photographers. According to South Carolina’s Joint Information Center, the state has not tracked data about whether new cases there are tied to the protests.

Meanwhile, data from other cities suggests protests have not been followed by an increase in cases of covid-19, the disease caused by the novel coronavirus. Minneapolis, where Floyd was killed and where the protests began, has registered a steady decrease in case numbers this month.

According to Minneapolis Department of Health spokesman Doug Schultz, more than 15,000 people were tested at centers the city set up in communities affected by the protests, and 1.7 percent of tests came back positive — below the statewide average of about 3.6 percent. Health systems in the area that tested thousands of people who attended the demonstrations returned positivity rates of less than 1 percent.

Schultz said officials believe the low infection rates reflect that the protests were outside, that most people wore masks and that people spent most of their time in motion, circulating through the crowd.

Officials in New York and Philadelphia have drawn similar conclusions and say they see no evidence of cases accumulating because of the protests.

In Seattle; Portland, Ore.; and Oakland, Calif. — cities experiencing a coronavirus resurgence — officials have asked people testing positive whether they attended protests, and few said they had. Neetu Balram, a spokeswoman for Alameda County, which includes Oakland, said officials there figured they would have identified signs of a demonstration-related outbreak by now but haven’t.

The same is true in Seattle. Out of more than 1,000 positive tests in recent weeks, 34 of the people testing positive said they attended a protest or mass gathering since late May, according to King County Health Officer Jeff Duchin. Nearly 3,000 people who said they were at protests have been tested by the city, and fewer than 1 percent were positive, Duchin said.

“The data may be imperfect, but we certainly don’t have any evidence that those gatherings outdoors are triggering this increase we’re seeing,” Duchin said.

In other places, the impact of the protests is less clear. Brent Andrew, spokesman for San Francisco’s Department of Public Health, said the city is still monitoring potential ties between a recent surge in cases and protests. In Houston, at the epicenter of a covid-19 crisis in Texas, officials attribute rising case loads to a variety of factors — and say protests could be one.

Houston Health Department spokesman Porfirio Villarreal said rising cases there could also reflect infections spread at Memorial Day gatherings and other family events, such as Mother’s Day and Father’s Day; graduations; bars where people failed to wear masks; and “people interpreting reopening as back to normal.”

Many states moved forward with reopening bars, restaurants, gyms and hair salons about the time the protests began. Some states — including Arizona, Florida and Texas — reopened as early as mid-May and were already seeing ominous trends before protests began. Surges in other states have also emerged 10 to 14 days after reopening — roughly the same time it can take someone who has been exposed to the virus to develop symptoms and undergo testing.

“You have many other things happening in states opening up. Really the only way, in my view, you can get a sense of where people get infected is through contact tracing,” said Jennifer Nuzzo, an epidemiologist at Johns Hopkins University Bloomberg School of Public Health, who added it is far easier to conduct contact tracing for a small gathering or family party than when tens of thousands of people pour into the streets.

Many of those participating in the protests were relatively young, Nuzzo said, and younger people are less likely to experience severe cases of covid-19 and therefore might be less likely to have symptoms that would prompt them to seek a test.

Still, available data suggests the protests did not ignite the same kind of outbreaks associated with some conferences, choir practices and religious services.

Many epidemiologists and virologists suggest being outside allows coronavirus-infected particles to disperse more easily, making outdoor gatherings — such as protests — less dangerous than those inside.

Angela L. Rasmussen, a virologist at Columbia University, said the difference in risk between indoor and outdoor settings can be explained by the way water moves through a fish tank as opposed to the open ocean.

In an indoor space, where air circulates through a limited area, people can be exposed to a higher concentration of respiratory droplets that may contain the virus. The more time people spend in that space, the more they will be exposed. Outside, while those droplets could reach others in the vicinity, they can also dissipate into the open air.

“While outdoor transmission is certainly possible, it does seem like it happens less frequently and that’s one of the reasons why: Your exposure is going to be higher indoors,” Rasmussen said.

Still, the relationship between being outside and exposure to the virus remains murky. Reports have surfaced of family barbecues or high school pool parties leading to major outbreaks, although it is unclear how much time people at those events might also have spent inside. Rasmussen said a variety of factors could explain why a small outdoor gathering spawns more covid-19 cases than a massive protest.

“If people are at a backyard barbecue, were they hanging out in the house together also? Were they in close proximity with each other? That would have an impact,” Rasmussen said. “I’d want to know if they were distancing, wearing masks, all of that.”

Economists from the Center for Health Economics and Policy Studies recently used anonymous cellphone data from the company Safeguard and covid-19 data from the Centers for Disease Control and Prevention to try to determine whether the protests spread the coronavirus.

The researchers explain in a preprint paper published by the National Bureau of Economic Research that they tracked more than 300 of the largest cities in the United States to see whether protests led to increased case numbers. They also used cellphone data to track social distancing in cities that held protests.

The paper found “no evidence that net COVID-19 case growth differentially rose following the onset of Black Lives Matter protests.”

The paper also concluded that, based on cellphone data, social distancing increased overall in cities that were home to protests — meaning so many more people stayed home in cities with protests that it canceled out the lack of social distancing by protesters.

But according to Dhaval Dave of Bentley University, an author of the paper, one shortcoming of the study is that it tracks covid-19 prevalence in a city’s entire population. In other words, the protests might have contributed to a rise in cases among certain demographic groups that didn’t manifest in broader data — once again blurring evidence and dulling the conclusions that can be drawn from it.

“Unfortunately, we live with the elephant sign philosophy,” Osterholm said. “I put a sign up in my front lawn three years ago to say ‘no elephants allowed.’ I have not had an elephant on my lawn in three years. So you think, see, it works!

“Epidemiology requires we think about much more than that.”