2011-11-14 16:25:02RW Wood and the Greenhouse Effect
Daniel Bailey
Daniel Bailey

The following is a guest-post by Eli Rabett:


RW Wood and the Greenhouse Effect

Posted on 21 November 2011 by EliRabett 

RW Wood was perhaps the greatest experimental physicist of his time.  Eli, old Rabett that he is did not know him, but he took classes from several who did and others who studied under him.  Wood was a practical man which gave his experimental work a particular clarity.

To the public, as far as he is known, he is famous for two experiments.  The first was his debunking of Blondiot's N-Rays, the second, his experiment purporting to show that the greenhouse effect did not happen as advertised. The Weasel has kindly posted the text of the later and Eli has had more than a passing interest in the same.  About a year ago he attempted a preliminary test of Wood's work, which was instructive, but not conclusive.  Rabett Labs was tooling up for a second improved try, when he was invited to submit something to Skeptical Science (cross-posted there and here, whichever there is there and here is here) and DeWitt Payne also wrote him to describe his attempt.

Let us briefly reprise what Wood did

To test the matter I constructed two enclosures of dead black cardboard, one covered with a glass plate, the other with a plate of rock-salt of equal thickness. The bulb of a themometer was inserted in each enclosure and the whole packed in cotton, with the exception of the transparent plates which were exposed. When exposed to sunlight the temperature rose gradually to 65 oC., the enclosure covered with the salt plate keeping a little ahead of the other, owing to the fact that it transmitted the longer waves from the sun, which were stopped by the glass. In order to eliminate this action the sunlight was first passed through a glass plate.  

There was now scarcely a difference of one degree between the temperatures of the two enclosure

There are several things wrong with this, well, not so much wrong, as not useful and to be complete, equally wrong with many so called demonstrations of the greenhouse effect that can be seen on You Tube.  To understand this the bunnies recommend a shortened version of one of Eli's simple greenhouse effect explanations, but boiled down

1.  Greenhouse gases absorption blocks some thermal radiation from the surface reaching space.

2.  Temperature decreases with altitude.

3.  The absorbed radiation and convective heating moves energy up from the surface

4.  The rate of emission to space in the blocked regions occurs from colder altitudes.

5.  Increased greenhouse gas concentrations mean that the altitude at which emission in the blocked regions occurs is higher and colder.

6.  Therefore energy loss is slower and the surface has to warm in order to restore radiative balance.

Now, let us look at Wood's experiment, first being McIntyre picky.  For one thing, the sensitivity of the thermometer is not impressive.  Also, in the case of the rock salt window a significant part of the IR from the black surface inside the cell is absorbed and for a third heating with sunlight brings additional problems.

the difference up to about 10 microns is due to reflections from the surface, and does not represent absorption.  Still, the major problem(s) are that

(a) the walls of the cell are excellent absorbers of thermal energy and light and (b) most importantly, there almost certainly was not a temperature difference between the top and the bottom of the cell.  We can get a hint of what this means by looking at some MODTRAN spectra.  In the first, looking down from 20 km

while if we look down from 1 km what the bunnies see is

because the temperature at the top and the bottom of the path are about the same and emission and absorption of the greenhouse gases is balanced.  In the case of the glass window, the IR emission came from the surface  of the window.  In the case of the rock salt, from the black walls which  were at the same temperature.  For the purpose of this analysis, the  emissivity of the two materials can be taken as equal, at least for the  sensitivity of the thermometer.

This tells us a few things about how to do a new and improved Wood experiment.  First, the inside of the cell should be silvered, to reflect IR light.  Second, heating should be done at the bottom of the cell.  Third, the top should be colder than the bottom.  Fourth, a much more sensitive thermometer is needed.  Fifth, KBr would be a better window than NaCl, but that is a quibble.

2011-11-14 18:54:14
Tom Curtis


This is an experiment I have given substantial thought to, of and on.  IMO a successful experiment would included the following features:


1)  A refrigerated glass screeen between the boxes and the sun to eliminate back radiation as a contaminant.  Alternatively the experiment could be performed using a high energy incandescent bulb as the heat source, with the experiment located in a refrigerated environment, or a very low emissivity environment for the same purpose;


2)  The glass or rock salt cover must be thermally isolated from the bottom of each box by using a material with a very low thermal conductivity to build the box, and ensuring the box itself is evacuated (no air).


With regard to Eli's points:

First, and contrary to Eli, the sides of the box need not be silvered so long as they are thermally isolated from the lid, and ideally are thermally coupled to the floor of the box by high conductivity material.

Second, while I have considered using a heat source at the base of the box, deniers would instantly seize on that as invalidating the experiment as a demonstration of the greenhouse effect.  They are wrong, but why buy trouble.

Third, we cannot ensure that the lid is cooler than the floor except by thermal isolation without contaminating the experiment.  In particular, any attempt to refrigerate the lid would just represent a secondary method for heat to escape, thus voiding the experiment.  On the other hand, if the lid is thermally isolated, it should approximate to a slab model of the greenhouse effect, and result in a temperature differential without further intervention.

I agree with Eli's fourth and fifth points.

Just because I have though hard about this experiment does not mean I have thought well, so I look forward to Eli's response (if any).


Finally, I have always thought that this sort of post is best done once you have done the experiment, and should include the youtube.  Absent that, denier's will just point to a number of failed attempts to replicate the experiment and argue the toss.  A successful experiment accompanied with an explanation as to why it succeeded while the others failed would be hard to argue.

2011-11-14 22:02:59


The central point of a simulated greenhouse effect is the presence of a thermal gradient. The simplest way I can think of to have one is the same as (iirc) the De Saussure experiment, i.e. having more than one pane of glass. This would be the analogue of the zero-dimensional model with more than one layer of fully absorbing atmosphere.

Experimentally, one may want to account for reflection losses by filtering the incoming sunlight with the same number of panes eventually pulled out from the device; one more advantage would be to avoid heating of the outer glass by IR from the sun.
I expect the effect to be large enough to be measured by standard mercury or alchool thermometers as De Saussure did.

As for Eli's post, it's ok to publish it as it is and have this discussion in the comment thread. If I find the time I'll calculate the effect with realistic parameters.

2011-11-14 22:14:06
Dikran Marsupial
Gavin Cawley

Might be worth noting that it was already known by the time of Fourier that the atmospheric greenhouse effect doesn't work the way a greenhouse works (so Woods and the scientist who recently confirmed Woods experiment, were really just refuting one of the earliest skeptic canards ;o)


The one change that is really necessary is "first being McIntyre picky".  SkS should avoid personal snarks at skeptics and stick to the science (to be honest, I'm not sure rabbettisms really work well outside Eli's normal burrow anyway -it will confuse those bunnies not used to that sort of thing).  Having said which, keep up the good work!

2011-11-14 22:58:11


Based on earlier comments:

- I suggest an introductory paragraph that gives SkS a little distance from this article: Make it clear that it is ER's point of view.

- Frame the article as a critique of possible problems with Wood's experimental set-up, rather than as "the right way to do it": Until the experiment has been done, you won't really know if you've thought of everything that can throw you off.

- Still, maybe ask him to curb the "rabettisms": It's just a little too-too, I think.

2011-11-16 12:23:12
Daniel Bailey
Daniel Bailey

Thanks for the thoughts.  I'm going to forward these to Eli tomorrow (in case he's been too busy to check in here); does anyone have any other observations/suggestions to make?

2011-11-16 12:38:43
Tom Curtis


As you specifically ask:

I think neal's idea of an introductory paragraph is a very good one, as also his suggestion about framing.

I disagree about the rabettisms, which I have always enjoyed.  So long as the introductory paragraph highlights that it is Eli's post, and Eli's style, which many bunnies (and anonymice) enjoy, there is no further problem with it.

2011-11-22 15:34:33


Thanks for the thoughts. Every time I touch this thing I learn something

We all agree that the greenhouse effect depends on the existence of a thermal gradient, so an experiment to illustrate the greenhouse effect has to have a substantial thermal gradient between the part that is heated and the part where the IR light is emitted to the outside.

To illustrate the greenhouse effect in a Wood's like experiment one would compare the internal temperature when the IR is blocked by a IR absorbing window or let out through an IR transparent window, where the windows are held at a temperature substantially below that of the bottom of the container. 

A simple illustrative experiment can be found at


where the temperature at the surface of a light bulb was measured with and without an aluminum foil cover.  The temperature of the glass envelope is a lot lower than that of the filament.  As Phil pointed out in the last comment

Or you can do what is done with some commercial quartz halogen lamps which is to coat the glass with a dichroic coating which reflects IR and transmits visible (aka 'hot' mirror). This causes the lamp to run hotter and increases the output (as I recall a 37W lamp has the same output as a 50W conventional lamp).

That operates in the near IR (under 3 microns) not in the thermal IR (~ 10 microns) but the principal is the same, the lamp w/o the IR coating lets more of the thermal energy out and runs cooler and is a great example of the greenhouse effect.

The major problem with Wood's experiment was that there was no thermal gradient, or at least a big enough one to measure with his thermometer. The black walls did not help especially since it is not clear that the light fell only on the bottom.  You want the heat source at the bottom, whether you heat by electrical resistance or by light absorption.

The lamp experiment OTOH has a ferrocious thermal gradient.  By inference not only are the envelopes hotter but so are the filaments (e.g. the surface of the earth).  If you want, you could heat the filament with a laser, or even a focussed white light beam (that would be a lot trickier).

A major problem in doing the "right" experiment is that you want to step clear of minor issues, for example, if you heat with absorption of light, you don't want to have to deal with the issue of a slightly different reflectivities of glass and salt windows for the incident light which is one of the reasons for heating from the bottom.

We went through the multiple pane thing in the reply to G&T.  It is one way of building a thermal gradient in.

More tomorrow.


2011-11-23 18:31:18Variation of transmissivity with wavelength is the 'greenhouse' effect
Larry Wade


Hi guys,

I've spent decades designing, building and using thermal systems/devices from over 1300 K down to 100 mK.  It may be that the language I'm used to is different than that used here.  In that case we may be saying the same thing but not communicating. 

It appears to me that your descriptions above are incorrect.  Here goes in the terms I'm used to:

First the context:  a hot object heats a mid-temperature object through radiative heat transfer.  The mid-temperature object is also thermally connected to a cold sink via radiative heat transfer.  The temperature of the mid-temperature object is stable when thermal equilibrium is established:  the radiative heat flow in is matched by the radiative heat flow out.

Lesson learned: the middle object has to be well insulated except for the two radiative thermal paths. The insulative value of cotton is not sufficient.  A vacuum is good.

The primary key to the greenhouse effect is a variation of transmissivity with wavelength (critically it is clear in the visible and gray to opaque in the IR). Our atmosphere, like glass, is highly transmissive in the visible and nearly opaque beyond a couple microns.  The energy distribution of light from a hot source is biased towards the short end of the spectrum. 

So the majority of the radiant energy passes through glass into a room or planet and heats it. 

If their is no 'glass' the room can readily radiate to space.  The room will then either warm or cool until the amount of energy radiated matches the amount of energy absorbed.

Now take that same room and add a plate of glass. If the glass (or atmosphere) is opaque or partially opaque to the radiated heat it will be absorbed at that layer and then reradiated (with 1/2 of the radiated energy being directed towards the room and 1/2 towards space). So now the room's effective radiant power is cut in half (for the opaque case).

The room will then warm from it's former equilibrium temperature. As it does so the intensity of the radiation (at all wavelengths) increases and radiation also occurs at shorter wavelengths.  As the room radiates more energy it in turn warms the glass (atmosphere). Eventually a new equilibrium is established with the room being warmer than it was without the glass being in the way.

Add another pane of glass and the effect continues...from this additional layer a gradient is established. So basically the radiant is not the cause of the green house effect but the instead is the result of it.

As more and more greenhouse gas is added to our atmosphere, the atmosphere absorbs and then re-emits energy: basically acting as additional glass panes.

Look up multi-layer insulation.  It works in a similar fashion.

To recap: thermal gradient is unimportant for a highly transmissive media because light passes in either direction regardless of wavelength.  The key is that our atmosphere is opaque in the IR and not in the visible.

Two additional small points: the peak emission wavelength for a 300K body is about 10 microns.  So a cooler object (our atmosphere) is radiating primarily at wavelengths longer than 10 microns. The other point is that gas is a pretty good insulator. Since 10 micron photons typically travels a fair distance before being absorbed the intervening gas can serve as an insulator...thereby enabling a thermal gradient to build up.

With the above in mind it is easy to design a relevant experiment. 

Put halogen lights in a vacuum chamber to simulate the sun.  Cool the walls of the chamber to ~80K.  Locate 4 objects in the chamber:  one black and bare, one black with a gas layer (for insulation) between it and a glass layer, the third adds one more gas layer (for insulation) and another glass layer. The fourth adds one more gas insulating layer with an outer glass layer.  You'll have no problem measuring the difference in the surface temperatures of the four balls.

2011-11-24 11:56:36
Dana Nuccitelli

Is this post ready to publish?

2011-11-24 13:08:07
Daniel Bailey
Daniel Bailey

From Eli's comment, dana, I don't quite think so.

I'll email him to see if he can give us a timeframe.

2011-11-24 17:40:37
Tom Curtis


Larry Wade, I believe your comments to by very apt to this case, and I like your experiment design.


Your comment about multi-layer insulation reminded me that everytime somebody double glazes a window, they are performing an analogue experiment, in their case retaining the same or greater heat with less energy produced by fuel (or electricity consumption).


With regard to the energy gradient, in these models they are indeed produced by the greenhouse effect.  However, in the natural greenhouse effect the energy gradient is generated by the effects of the ideal gas law in a gravitational field.  I believe that this is what Eli has in mind, but I do not believe it can be reproduced in small scale, simple experiments.  The best we can do is reproduce the slab atmosphere models of the greenhouse effect.

2011-11-25 11:40:58


Would it not be pertinent to point out that the experiment (because of its design) does not disprove the greenhouse effect but does hint that it is poorly named? Ie the heating in a greenhouse is not caused by greenhouse gases.

2011-12-06 15:41:49


I think this post could use (but does not require) a simple schematic of the experimental setup, and if possible also representations of what Wood imagined was going on and what it actually gonig on (light in, IR emitted, convection, relative temperatures at various locations in the setup).

I have it all straight in my head (or so I think) after having read a number of writeups on it, including all of the comments in Eli's own blog, but... I think for many readers it takes a lot of hard thinking to put all of the pieces together.  There may be a lot of confusion for people who haven't seen Wood's experiment before.

2011-12-06 18:12:38
Glenn Tamblyn




The emissions curves above don't make sense to me, unless I am missing something. That the 20Km curve has all that cool structure while the 1 km curve doesn't. Get that. But in both cases the atmosphere curve is significantly higher than the 300 K curve. Huh? How hot is the atmosphere? What am I missing?

2011-12-07 09:17:30



I think you're right.  I generated the two (20k, 1k) from here (really easy to do):

Modtran run me

and got pretty much what you'd expect, like the images above but with the red ATM line properly placed. 

2011-12-17 21:12:13
Tom Curtis


If this blog is advanced further, this paper by Wagoner, Liue and Tobin (2010) would be a useful resource. (h/t to skept.fr)

2011-12-18 02:43:30
Daniel Bailey
Daniel Bailey

Just emailed Eli.

2012-01-08 06:07:04FYI
Daniel Bailey
Daniel Bailey

Eli & Josh are working on this (just emailed me to let me know) again.

ETA unknown