Skip to Main Content

Speaking calmly and at a normal volume produces liquid droplets so small they can remain suspended in the air long enough to enter the airways of other people, potentially exposing them to viruses including the one that causes Covid-19, according to a new study led by scientists at the National Institutes of Health.

“Aerosols from infected persons may therefore pose an inhalation threat even at considerable distances and in enclosed spaces, particularly if there is poor ventilation,” Harvard University biologist Matthew Meselson wrote in a commentary accompanying the paper, which used a laser to visualize airborne droplets created when volunteers uttered the words “stay healthy.”

The study was published on Wednesday in the New England Journal of Medicine.


The question of whether the coronavirus can be “aerosolized” has stirred controversy for weeks, with a study last month reporting that the virus may be able to stay suspended in air under special circumstances, such as when infected people undergo intubation and other procedures. But the new study suggests that the SARS-CoV-2 coronavirus can be transmitted from one person to another simply by speaking.

Large particles such as those expelled in a sneeze or cough “remain airborne only briefly before settling because of gravity,” Meselson wrote. But “breathing and talking also produce smaller and much more numerous particles” that are “too small to settle.” These aerosols are therefore carried by air currents as mild as those generated by people walking around a room, drafts from open windows and doors, and vents that create air flows.


The large droplets that settle on surfaces can cause infection if people touch them and then touch their face. In that case, the droplets and any virus they carry are deposited in the upper respiratory tract, where with any luck they’re removed in nasal secretions or swallowed before they cause an infection.

The aerosols, in contrast, are so small (a few microns, or a couple ten-thousandths of an inch) that they penetrate deep into the lungs, infecting cells.

For their study, scientists led by Philip Anfinrud and Adriaan Bax of the National Institutes of Health asked volunteers to say “stay healthy” into the open end of a cardboard box whose inside was painted black. They used a green laser to create a sheet of light three inches from the open end that, after the person spoke, captured any droplets that reached it. An iPhone 11 Pro video camera recorded the arriving droplets, which produced flashes as they passed through the laser light sheet, allowing the scientists to estimate their size.

“Numerous [aerosol] droplets … were generated,” the scientists reported. The “th” sound of the word healthy produced the most droplets, and speaking loudly but still in a conversational voice produced more droplets (347) than speaking softly (227).

When people said “stay healthy” through a slightly damp wash cloth, however, the number of droplets reaching the box fell to zero. The scientists did not measure how far the droplets could carry, and remain suspended in the air, under different environmental conditions, and no viruses were used in the experiment. But the earlier NEJM study suggests that droplets containing the coronavirus can become aerosolized.

Previous studies, unrelated to Covid-19, have found that droplets emitted during speech are smaller than those from coughing or sneezing and therefore more likely to hang in the air. But they do not do so forever. In still air, a particle with a diameter of 10 microns (roughly one-seventh the width of a human hair) remains aloft for roughly 9 minutes.

With aerosols now back on the table as a potential source of Covid-19 infection, Meselson wrote, it “suggests the advisability of wearing a suitable mask whenever it is thought that infected persons may be nearby and of providing adequate ventilation of enclosed spaces where such persons are known to be or may recently have been.”

As for who “such persons” might be, that’s often impossible to know: A study in Nature Medicine, also published on Wednesday, found that people with Covid-19 are infectious two to three days before they show symptoms.

  • This is very late confirmation of the very logical act of wearing a mask. This barrier reduces the aerosol effect, not eliminate, but it does provide increased protection. Unbelievable, that the wearing of masks was first advised against – as ineffective (obviously wrong) – or stigmatizing (who cares ???). If everyone makes it a habit to wear one when not under one’s own roof, thousands of lives will be saved.

  • Studies have been done weeks months ago, on aerosolation and its reach, its depersion with air movement etc etc. OF COURSE speaking produces aerosol – so does simply breathing. Therefore the distance now behind a jogger = 10 meter and behind a biker = 20 meter. I am stunned that it has taken well over 3 months to figure out this so very logical spread. People are needlessly dying because of this to me retarded delay.

  • “Potentially”, but in practice it seems this is not causing many infections, except prolonged in confined spaces. If they did, contact tracing could not work. There would be untraceable infected people everywhere.

  • This is not an all or none issue. Of course there is some aerosolization. The question is the amount. If 10 people on an elevator are all infected and talking, perhaps this amount becomes clinically significant. You must follow the epidemiology to understand the primary mode of infection. This suggests droplets. Pre-symptomatic infection does not prove aerosolization is primary. I suspect those infections would be close contacts.

    • I am not disputing your Comment is true generally, but i am not sure we are able to follow the epidemiology very well. I guess i feel, if I am going to be locked down so that everywhere I go is voluntary, I want to know about any risk from other people – if there is any at all, then I can choose to avoid it.
      The other thing, not to get into politics, but I have felt we could have gotten more for our “money” in the lockdown.
      Meaning, the economic cost of the lockdown was say, $500B, and for that, short term, we got a few million fewer sick people – but, one of the assumptions in the all the public health advice was no airborne transmission further than 6 feet, therefore no need for face masks if you keep the 6 feet separation – but now it appears fairly likely a lot of people were infected while following all the lockdown rules – and while the person infecting them was also following all the rules. Simply making rules to wear face masks, even homemade face masks, would have possibly cut down the epidemic much more, for the same “cost” in lost productivity.

  • I am going to attempt to paste in a link to an interview with Dr. Kim Woo-Ju, who is one of the top experts on infectious disease in South Korea. In this fascinating discussion from several weeks ago he discusses how this virus is easily spread just by talking (he references some laser light research). In addition, he talks about their (Korea’s) largest super spreader event which was a church service where people were singing and shouting in a large crowded room. A perfect environment for spreading respiratory virus, according to the Dr. Also, mask wearing is nearly universal when out in public in Korea right now.

    • Thank you for the link. I saw a program on Australian 60 Minutes on YouTube in which they interviewed Gabriel Leung, a prominent virologist in Hong Kong, who said masks were very protective. I never went to other parts of Asia, but visiting Japan and seeing Asians here, I was aware there is much greater mask wearing. The thing about the N95 masks, if CDC was going to lie and tell us face masks would not protect us, maybe they should have changed the lie to the claim the surgical masks were sufficient, so that infected people might have worn something which would have protected other people. Going to Costco and seeing (two weeks after the 6 feet recommendation came out) a neat line of people separated by 6 feet – but with the line pointed directly upwind, so that everyone in line was getting all the air from all the people ahead of them in line, made me realize I needed to try to research this myself.

  • The article neglected to emphasize that the distance of the detection from the speaker was not clear in the summary, but somewhere between 18 and 24 inches.

    Also, “We did not assess the relative roles of droplets generated during speech, droplet nuclei,2 and aerosols in the transmission of viruses. Our aim was to provide visual evidence of speech-generated droplets”

    In other words, this is a very limited study that show that a small amount of droplets and aerosols distribute about 1.5 feet from someone’s face when they talk, but says nothing about how likely this is to infect someone in the real world. Guidance for social distancing is already 6 feet, making this practically irrelevant and not new information. This is just like other coronaviruses.

    Then in the comment section we get insanity claiming everyone needs to wear n95’s and helmets at all times. No one recommends wearing n95’s all the time, it is nearly impossible to put up with the discomfort for long periods, and unhealthily suppresses your breathing and oxygen intake, leading even to distorted blood markers.

    This is a perfect example of why people who are untrained in interpreting studies and haven’t spent years digesting and contextualizing the literature should not be playing armchair scientist with individual studies.

  • I am not sure why the study chooses to use a damp washcloth to block transmission. Those are not practical for people to use. It would have been nice for them to use a bandana to show whether it helps at all.

    The reality that we all need N95 masks to inhibit transmission has been evident for a while and these guidelines to use a cloth mask are seemingly laughable. What chance does a bandana have against microparticles.

  • Of course. If we seek to quantify risk to this degree of minutia; then yes, there is no absolute save vs. unsafe value. Although, it is in our nature to seek absolutes; to quantify what is required to guarantee safe, and what exactly makes it unsafe. If we “quantify” every single death as unacceptable; then this research demonstrates that we simply must not speak.

    • Although this is scare because some fool insisted on getting near me – I did not see him in time to drive him away – I want to know, personally. You actually have no basis for claiming the risk is “minute”. This does bring up the need to get good full head helmets, like a motorcycle helmet, with good N95 mask underneath, before getting near anyone, and I am glad that is being pointed out.

Comments are closed.