Public Lab Research note

  • 2

A kit to design your own formaldehyde experiment?

by warren |

I've been mulling over the amazing work @nshapiro and others have been doing on formaldehyde detection and remediation -- especially the plant-based remediation kit folks have been testing out -- and had a few ideas I wanted to share for input.

I was thinking about whether it'd be possible to set up a version of the plant based filter, as shown here, but with some tweaks, and I'll get to why in a moment:

image description

In this prototype, it's hard to structure a blind test, where you aren't "primed" by knowing that the filter is running, and so you may be feeling a placebo effect (though really, any reduction in symptoms would be great). I wanted to think of a way around this, and was inspired by the NASAL RANGER blogged about by Nadya Peek:

image description

Self-blind testing

The interesting part about the NASAL RANGER is that it lets you flip on and off the valve drawing in the smell you're pointing it at. That is, you can compare between the smell and the absence of the smell.

But I thought -- what if that were randomized when you pressed the button? If you could find out only after your ran the test whether you were smelling it? Or likewise, if you could be running the plant air filter OR NOT but could tell retrospectively afterwards? It could enable you to run a blind test, where you record your symptoms over a period of time, and then find out whether any change (for better or worse) correlated with whether the filter had been on.

So, I thought, maybe you could just run the pump but shut off the valve. But just to think a little further, I imagined a bell jar (like the one the Little Prince used to protect his rose, or the one in Beauty and the Beast) with a kind of black box as a base, which had two fans inside. One fan would pull air thorough the glass chamber, and the other would just bypass the chamber, but you couldn't tell which was actually running -- both would cause some air to flow out the exhaust vent.

image description

So, you'd run it for a period of time -- I'm not sure what that'd be, a week? A month? I'm sure @nshapiro would know better. And you'd know that it might or might not be on. But when you're done, you turn it off, and it shows (with, say, a blinking light or something) whether it had been on for real or not. You add this to your journals, and you've conducted a self-study. You can repeat it, change things, etc etc, to modify the study.

Refining the study, not the sensor

One thing I really like about this is that instead of the kit being the sensing technique, it's the study design. You can change the period you observe over, you could run it in different sized rooms, or track different symptoms. You can choose to do a standardized test that many other people with this kit are doing, so you could compare, or you could branch off and try a different variation.

It also seems powerful that, should you choose to share your data, you're sharing a study you've run yourself, rather than a researcher knowing if it's a placebo or not (I'm reminded of Alan Irwin's 1995 case study on citizen science by AIDS activists who argued that placebo-based blind studies were unethical, and -- do I have this right? -- swapped pills to "take over" the studies). Of course, such a kit could also have a switch so that it's never a placebo -- that's in the hands of the individual. They could just run the study themselves and only compare it privately with others' published results, without posting their own. Or they could decide to post their data and co-create a broader study with others.

This is especially interesting to me because I think community environmental science tends to focus a lot on the sensor -- and very narrowly -- often just the electronics, to the neglect of things like study design, setup, and issues of privacy, control, and feedback:


One thing I like a lot about doing a remediation kit is that, unlike so many environmental testing projects, it involves a feedback loop. Instead of just putting a sensor or something in your window, and learning something about the presence of pollutants that you may not be able to substantively improve, you're actually changing the space you live in. And you're also measuring it to see if you've changed it.

The feedback loop is one reason I think folks can get really addicted to FitBits and the kind of self-tracking from the Quantified Self movement -- change something, and see its effects. So I wanted to think about how the remediation kit and a means of measuring formaldehyde could interact.


I didn't want to get to far ahead of myself, but I did look into some parts and prices. There are 1 gallon and 2.5 gallon glass fermentation jars with standard 4" lids available online:

image description image description

They also, interestingly, often have holes in them for valves. I think a bigger hole would be needed, especially if I'm going to look at how airflow with small fans could work -- is it consistent enough? I do think it'd be a lot cheaper and more compact, as well as quieter. But it could work either way.

Anyhow, that's a lot -- I'd be glad to hear peoples' thoughts, and may want to try making some prototype "placebo jars" if anyone's interested in using them!

remediation formaldehyde indoor-air-quality prototyping indoor-air placebo study-design feedback-loop priming blind-testing double-blind


I managed to cobble together an Arduino program to try this out:

Button turns on one of two fans, randomly. On second press, it reveals which had been on by turning on one of two corresponding LEDs.

And a circuit (will post in a moment) that fits in a jar lid -- although I need to plug in both fans, and need a couple transistors since I guess the Arduino can't drive these fans directly from its pins (I'm still a newbie after all these years using Arduinos...).

It'll need some ducting design, and i guess a tube to go down into the roots from the lid. But it's a start!


I love this idea! It really highlights that this kind of intervention could be a way to make the diffuse impacts of indoor air quality quantifiable in a robust way, but without having to quantify which toxins are present (although we can do that as well). These symptoms have been remarkably hard to tie to homes and toxicants. Knowing what plants and growth media foster which bacteria which degrade which toxicants will be important for making the results of this study even stronger.

It's a whole kind of study design that is almost more like pharmaceutical trials. I wonder if we could do this in an analog way where we just puncture the diaphragm of a pump so it goes put pumps no air.

I'm not sure those fans will be able to move enough air to sufficiently diffuse the roots. The Andrea (pictured below) tried to do exactly that and consumer reports found that it really didn't do what it was supposed to. That's why i'm such a fan of the fish pump.


I wonder if there is a possibility of an arduino that is between the pump and the plant that allows air to go into the plant root or not, after the occupant logs their symptoms online every week. this is one way of doing randomization without massive scale called the N of 1 trial, an often overlooked model.

I think the ethical considerations are really important here, thank you for underlining those. Having a quick way to turn on the placebo devices should we find them to be conclusively effective seems like a vital consideration.

What do we need to do to help move this idea closer to reality?

Hi, Nick, thanks! Using an arduino to switch pumps or valves seems totally plausible. For prototyping, fans was easier, and I do wonder if we could:

  1. look up flow rates and specs for fans and pumps
  2. emprically measure flow with both (like how @mathew did with the minibuck)

How did Consumer Reports measure this? But you're right, we should probably first try to get this running with pumps.


  1. Is a 1 or 2 gallon jar big enough to do a test with a week-long period?
  2. Are there any smaller pumps that would still work?


I think we could use adapters or something there... but do we need a seal that tight? For air, could we just flip a flap with a servo or something? But it'd be audible... hmm.

1) Looking up the flow rate is going to give a rough estimate, but we will probably have to measure in-situ. The bubble method is a good one here. We can also use plastic bags of a specific volume and fill them up, like 1-gallon ziplock bags.

just opening/closing the valve could potentially damage the pump. The pump also tends to whine in distress when closed off, and so it would be pretty noticeable. I think you'd want a bi-directional control valve that would dump air from the pump before getting to the filter media. There should be '3-way' or 3-port valves available where they usually empty out an exhaust port but can be powered to send flow in a chosen direction. I don't have an example in front of me, but that should solve the sound issues.

yes! a 3 way would be great, they really do moan when they get jammed. A three way or a two way that vents the non-line engaged channel would be ideal.

Here is a wayback machine link to the Consumer Report:

They claim it didn't perform well and that the makers misrepresented a third party study in their promotional materials.

Measuring in situ would be great as the volume that a fan can do might be deceptively enticing when pulling air through the growth medium. They may have less force than the aquarium pumps?

in terms of jeff's questions: I'm not sure I understand the bell jar totally. I think a month is a good time for a test, to minimize illusory short-term changes. i think @zengirl2 might have some nice ideas about housing. and i think @kkoerner did some nice research on other pumps

Also funnily enough the guy who is quoted in the consumer report is a co-PI on our NSF grant. (sorry for haste, this is written from the airport)

I do have some documents and quotes for smaller pumps, though they're more costly (roughly $100 with shipping included) and when LukeS tested similar pumps they got mixed results. These were mostly to save shipping weight/size for the formaldehyde test kit, as the aquarium pump is a little bulky.

It looks like there's some more research on this here. If you can find a cheap 3 diaphragm miniature DC pump, that may work as well as the aquarium pump.

This is all very interesting @warren and @nshapiro because initially on reading the report with the air device and hearing the user's symptoms had improved, I wondered about the accuracy. So, the idea of improving the experiment itself is great. As far as containers, I have been collecting notes on Pinterest for waterproof household type items for garden plantings. These range from boots (yes, boots!) to pouches and everything in-between. Nick had mentioned some of the buy-in for the device was based on looks. So, I started thinking of ways to help hide all the tubing and pump, as well as eliminate the need to put a hole in the container. So far my fave method would be a container in a container. The inner container would be a practical easy-to-find item like a 64 oz. plastic soda bottle that has been cut down. It could be placed in another larger container to house the pump, like a basket or larger ceramic pot. When Nick and I discussed this, certainly one of the issues is adding more toxins to the equation like plastic, however, like the Riffle, plastic is easy to obtain and can be cut down to create the correct size needed. Having people supply their own container also eliminates the need to ship it in kit form. So, I will work on putting together a field note with pics.

You must be logged in to comment.