Updated: Jun 11, 2021
It is now acknowledged that the dominant means of COVID-19 transmission is not simply airborne transmission, but specifically airborne transmission of respiratory aerosols exhaled during normal breathing. Embracing this ought to have had a profound impact on the measures we use to mitigate the spread of COVID-19, specifically masks. It has not.
The reason this is so important, is we KNOW the size distribution of respiratory aerosols--we have since the 80's. We also know how they can--and cannot--be filtered. They range from ~100 nm - 1 micron, with more than half smaller than 0.28 microns (280 nm), and the majority between 0.2 and 0.5 microns--a range that has a special name called "Most Penetrating Particles," (MPPs), that should give you an idea of the challenge. This size distribution is nearly identical to that of cigarette smoke, where the majority of particles cluster around the 0.2 – 0.25 micron range. Fewer than 0.1% of respiratory aerosols are above 5 microns. It is likewise known that particles over 5 microns carry less than 0.1% of virus (not a coincidence). Beyond that, multiple studies for flu and COVID have shown that when 5+ micron aerosols carry virus, that virus is non-replicating and non-infectious.
Figure 1: Size Distribution of Respiratory Aerosols Generated During Tidal Breathing (Healthy Human)
The importance of the graph above cannot be over-stated. People struggle to understand why masks might not “work” to stop COVID and other respiratory diseases, because they continue to think of the virus as being carried by tiny droplets of spit that should be trapped by the masks. If, instead, people were to think of the virus as being carried on invisible aerosols that are the size of the particles in cigarette smoke (which they are), and which, by virtue of their size, behave just like smoke, the problem would become much clearer.
Given that cigarette smoke is the same size as the respiratory aerosols that carry 99.9% of the exhaled virus, exhaling cigarette smoke with various masks provides us the ability to easily visualize the efficacy of different masks in containing these aerosols. By the same token, the ability to smell cigarette smoke through a mask allows a handy trick to test the efficacy of masks in protecting users from SARS-CoV-2, flu, and other viruses carried in respiratory aerosols (indeed, the "smell test" is one of the qualitative ways that N-95's are fit-tested).
The video below shows me exhaling smoke through a variety of different masks, including:
Ubiquitous double-layer cloth mask,
Surgical mask with the earloops crossed, and the sides tucked
KN-95 with a cloth mask over it to improve fit
Fitted N-95 surgical respirator
Fitted N-95 medical respirator
Fitted N-95 particulate respirator with the exhale valve open
Fitted N-95 particulate respirator with the exhale valve taped closed
KN-95 with a fabric mask over the top, and a gaitor on top of that, pulled all the way over my head
A demonstration of an attempt to blow out a match with a double-mask both in front of the mask (not possible), and holding the match above where my breath escapes (easily done)
Video 1: Exhaling Cigarette Smoke Through Various Masks
In all of the examples, you can see that the smoke easily escapes through the sides or the top of the mask by the bridge of my nose, despite my making an effort to create a proper fit. Only in the case of the fabric mask does the smoke visibly pass through the mask. For all other scenarios, the mask acts not like a net that catches the particles, but rather like a wall that redirects the air flow of my exhalation.
This is a physics problem, not a biology problem. The determinants for filtration efficiency have everything to do with particle size and fluid dynamics. Virtually every single study claiming that “masks work” to stop aerosols in this size range bases that claim on the filtration efficiency of the fabric—they studiously avoid acknowledging the importance of the gaps. And yet the gap is all-important. It is for this reason that the concept of fit-testing for N-95 exists—to eliminate the gaps. As you can see in Table 1 below, even for an N-95, a 1% gap renders them only 12-35% efficient (depending on the particle size).
N-95s are not designed to keep particles IN, they are designed to keep particles OUT. The act of exhaling alone, even in a fit-tested N-95, creates its own small gaps, as the force of the exhalation pushes it away from the face. The smoke (or respiratory aerosols) then escapes through those gaps. By contrast, the negative pressure created by inhalation serves to tighten the seal, and improve filtration in N-95s, improving the efficacy for their stated purpose--protection from airborne particles over 300 nm.
For a fabric or surgical mask, as is shown below, the degradation of performance goes to zero with just a 3.2% gap—that is what you saw in those videos. The smoke going out the front of the mask was what was “filtered." The rest--the vast majority--simply escaped, as air will. Again, to put this in perspective, a 1 mm gap along the sides of the mask is enough to create a 1% gap. The gaps in un-fitted masks are typically on the order of ¼” wide--making these gaps much larger than the 3.2% gaps that render cloth and surgical masks completely ineffective. While we're here, let's stop and remember who in our society remains masked--only children and retail workers--both of whom tend to have only the lowest quality fabric masks with large gaps. The masks these two groups wear, and the way they wear them--loose, with large gaps, would do nothing to protect anyone, even if they were N-95s. It seems odd to me that teachers feel safer with kids wearing masks. Watching the videos, it seems clear that the only thing these masks do is to concentrate the aerosols jet and send it up--often times presumably right into a teacher's face.
Source: https://www.tandfonline.com/doi/full/10.1080/02786826.2020.1817846 Extrapolation, courtesy, Stephen Petty, image courtesy of Daniel Horowitz.
One source I found via the CDC actually pointed this out to me very elegantly. In the study (which was a Ph.D. thesis) the author compared the filtration efficiency of 3 surgical masks, applied “normally”, and 3 surgical masks that were literally glued with silicon to the face of the mannequins to eliminate the gaps.
This particular study was looking not at the masks’ ability to keep particles IN, but rather, their ability to keep 0.5 micron particles OUT (protection, vs. source control--again, what masks have always been designed for in the past). In the study, the 3 masks that were not glued to the face had an average filtration of 20%, ranging from 3.8% to 43%. The study doesn’t indicate the reason for the variability, but gap size seems like a reasonable guess, particularly as the masks that were glued to the mannequin showed an efficiency very similar to that of the most effective of the “unsealed” masks--which were between 43% and 51%.
There are two additional things to note here. First, this study looked at particles that were uniformly sized at 0.5 microns. The majority of respiratory aerosols are 0.28 microns or below—substantially smaller. The other thing to note is, again, this was testing protection (keeping things out), not source control (keeping things in). Masks are far better suited at keeping things out (even surgical masks) than keeping them in, because the force of exhalation continually pushes particles out. What this tells us, is that even with a perfect seal, with particles that are twice the size of the respiratory aerosols that carry the 99.9% of COVID and other viruses, surgical masks only provide at a maximum, 50% protection. In a real-world situation, they provide closer to 20%--for 3 minutes. But again, the respiratory aerosols carrying COVID are less than half that size, so reduce it even further. Then recall that for capture the efficiency is even less—as mapped in excerpt three above.
A very elegant study was done way back in 1980 the illustrates this concept well. In order to test the efficiency of surgical masks at preventing wound contamination from bacterial infection (that is their purpose), a tracer particle was placed on the INSIDE of a surgical mask. In every single test (20 in total), the particle ended up IN the wound.
This finding was echoed in the 2016 study, below, which found no impact on wound infection due to surgeons wearing masks.
While gaps tell most of the story, they don't tell the whole story. Cigarette smoke is a useful tool because it allows us to easily visualize the behavior of these sub-microscopic aerosols. But only for a few seconds. Once they disperse, they are no longer visible--but that doesn't mean they are no longer present. A smell test is enough to make that clear.
The smoke that escapes through the sides and top of the N-95 masks is not the only smoke that escapes--it's just the most visible. Other smoke escapes through the gaps as it is pushed out in subsequent exhalations, or as it passes through the filter itself. It takes more time--say 3-5 seconds, rather than 0.5 seconds--but it still escapes. well. Yes, the valved N-95 creates a large concentrated jet of respiratory aerosols. But there are still plenty of respiratory aerosols exiting the un-valved N-95, both through gaps, as well through the mask itself (see video).
Respiratory Aerosols Escaping Un-Valved N-95s
Video: Respiratory Aerosols Escaping Valved and Un-Valved N-95s
Respiratory Aerosols Escaping Un-Valved N-95s
Once again, this squares with other observations. It is worth remembering that even without any gap at all, N-95's are only supposed to stop 95% of particles GREATER than 0.3 microns--on inhalation. Remember, more than half of respiratory aerosols fall below 0.3 microns for a healthy person (closer to 90% in an ill person). In graph A of Excerpt 8 below, we observe that N-95s have their lowest filtration efficiency--even without a gap--in the exact range of respiratory aerosols, dropping down below 80%, with very large error bars.
The particle size distributions here apply to healthy people. A recent article examined the aerosol production of COVID-infected monkeys. What they found was truly stunning.
On day 7 after infection, the smallest aerosols, those that are 0.3 microns and below, increase by more than 7 fold, from around 10,000/5 minutes when healthy, to more than 70,000/5 minutes. This means, that when viral load is highest, production of the most infectious aerosols is also at its highest. At the same time, the aerosols that can be stopped by masks, those over 1 micron, decrease from 25 particles out of 11,000+ (0.23%), to FOUR out of 75,000+--or 0.005%.
Figure 2: Respiratory Aerosols by Size, Day of Infection, and Viral Load
What this tells us is that when a person is ill, masks are known to be effective stopping the aerosols that make up just 0.005% of exhaled aerosols. We likewise know that this size of aerosols, carries <0.1% of all virus. In fact, the amount is likewise even smaller, because the study that identified this did not look at aerosols smaller than 0.3 microns, which, when a person is ill, appear to make up more than 90% of all aerosols.
Thus, when it comes to the masks that most of us are wearing--even those wearing double masks--the argument "even if it helps just a little," carries no water--or at least stops no virus. We are literally talking about stopping fewer than 0.01% of the aerosols, which also carry less than 0.1% (probably more like 0.01%) of virus. Furthermore, Excerpt 10 shows that the <0.1% of particles that masks are able to stop, are also the particles that fall to the ground most quickly. A 0.2 micron particle remains suspended in the air nearly 600 times longer than a 5 micron particle. This is clearly one other thing contributing to the greater infectiousness of these particles.
Source: https://www.liebertpub.com/doi/full/10.1089/jamp.2020.1616 (time to settle 2 meters, was calculated based on the table--but was not included originally)
So what does it mean to remove <0.1% of the exhaled particles that stay in the air for 0.2% of the duration of the other 99.9% in terms of your likelihood of passing on your illness to someone else? Not a hell of a lot, but let's keep digging. For instance, how many viral copies does it take to actually get you sick, and could removing this tiny, most ephemeral fraction help reduce the likelihood of getting others sick?
Figure 3 attempts to quantify this, for various modes of breathing. If we think of infectiousness as the number of particles multiplied by the amount of time each particle suspended in the air, we see just how little impact removing larger particles have. For every single mode of exhalation, if we remove 100% of particles over 1 micron (this would assume that they were ALL stopped by impaction, and none escaped through gaps--this is not what happens in reality), we would still only reduce the infectiousness of a given exhalation by 0.1% to - 0.2%. Particles under 1 micron are not subject to filtration via impaction (running into stuff).
Now let's consider an N-95 mask. N-95 mask is designed to remove 95% of particles over 0.3 microns (upon inhalation--not exhalation). For this exercise, let's pretend it removed 100%. Let's also pretend that 100% of those particles between 0.3 and 1 micron were "caught" (again, this doesn't actually work for particles of this size, but let's pretend) in the masks, and didn't escape via the gaps. In this case, we would only have the particles that were less than 0.3 microns. Unfortunately, these particles are still more than half of all particles, and take weeks to settle. Thus, the overall "infectiousness quotient" is still only reduced by 9-11%. Interestingly, this is inline with what was seen above in terms of the effectiveness of an N-95 with a 1 mm gap.
The table below puts the numbers above into an easier to understand format.
Sources the same as figure 3
In the MIT study cited at the beginning of this study, the authors estimate that approximately 10 viral copies are necessary to cause infection. This would make COVID 30-300 times more infectious than flu, which seems like a stretch, but lets go with it.
How many viral copies are contained in these aerosols, and how many of the aerosols contain virus? This 2017 study found viral influenza RNA in 76% of fine aerosol samples, generating roughly 38,000 viral copies in a half hour. 40% of the samples had infectious virus. But someone only needs to be infected with 10 to get sick--and the most infectious particles likely to reach deepest into the lungs are likely to stay aloft for weeks. Using masks to try and stop this problem is akin to is using a nylon sock as a condom--with equally predictable results.
Many people will still argue, even if it's not effective, what's the harm? It makes people feel better, so let's just do it. Well, for one thing, one must consider the harm that arises from telling someone who is at risk of this disease, that they are protected by wearing a mask or by others masking, when in fact they are not--what does that do to their risk? What would it do to birth rates to tell people nylon stockings could be used as condoms to prevent pregnancy? COVID-19 is a very deadly disease for a very specific portion of our population. I have seen how this played out in real life, actually. Having heard of the CDC's case study that it was safe to get your hair cut, so long as every one wore a mask, my stepmother (59), got her hair cut. Both she and the stylist wore KN-95 masks the entire time. The day after the haircut, my stepmother got a call from her stylist telling her she had tested positive for COVID-19, and was now symptomatic. A few days later, my stepmother also tested positive. My stepmother was fine, but what if this were someone who was really at-risk? This erroneous guidance would have put her in true danger. The more specific we can be, the more tailored we can make the guidance to allow at-risk people to make smart decisions, and assess their own matrix of risks and benefits. The nature of respiratory aerosols makes this a challenge, but giving at least some guidance that is honest can help people make better decisions.
The next post will be focused on the many other harms that arise from universal masking.
On the other hand, if you doubt the above, you can listen to Dr. Michael Osterholm explain the same in June of 2020, below. Dr. Osterholm has since been forced to retract this--not on scientific grounds--but it is an eloquent and easy-to-understand explanation squaring with everything you read above.