Thread by @CappaSnappa

We tested neck gaiter performance for reducing exhaled particles when worn by a person speaking...results differ quite a bit from a conventional test using manikins (see

https://twitter.com/linseymarr/status/1294813717373345798?s=20

). Why? Read on (it's not droplet shattering #gaitergate) 1/n

When speaking the Rainbow Passage, we observe more particles when wearing this polyester gaiter than without (method: see Asadi et al. 2019/2020). But is this droplet shattering? To test this we "fake read" the same passage but without making any noise & breathing very little 2/n

The result: very few particles with no mask but a ton with the mask (single or double layer). The friction from face/gaiter rubbing seems to be generating particles from the mask. What's that you say? It could still be droplet shattering? 3/n

Nope! Next we tried holding our breath while simulating chewing. The result: almost no particles with no mask but substantial particles with the mask + holding breath. And much more compared to mouth breathing. 4/n

But what about that large particle mode? Friction! Compare what we get when we just rub the gaiter material against itself. Both small and large modes. Fabrics can shed! (Skin too, but that's another story.) 5/n

But why more when "fake" talking than when just holding breath + motion? Likely b/c the airflow helps carry particles to the detector. Why no large mode when talking? My guess is higher velocities (from air motion) enhance impaction & inhalation losses. 6/n

Is this unique to gaiters? ABSOLUTELY NOT. Different materials have different propensities to shed, and even for the same material (e.g. polyester or cotton) differences occur depending on the weave, thread thickness, etc. (a story for another day). 7/n

Does this mean gaiters are poor masks? PROBABLY NOT. This is where experiments informing material efficiencies, or with less mechanical stimulation (aka friction) can help, such as those by @isjinpan and @linseymarr. If you can't see the sun through it, it's probably ok 8/n

For more on this look for the (hopefully) soon to be out new paper by Asadi et al. with my colleagues @ucdavis and @MtSinaiQueens