A Creationist ‘Physicist’ is Confused
A Critique of Russ Humphreys’s AiG article on superluminal light propagation
This is a response to an article written by Russ Humphreys and published on the Answers in Genesis website about an experiment reported in Nature in which causality appeared to be violated and light seemed to arrive at a detector much sooner than the speed of light would allow. Humphreys's article is still on the Answers in Genesis website here:
Because Humphreys is seen by the creationist community as some sort of expert in physics, I believe it is worth spending time and effort on this as it does bring his rather serious limitations as a commentator on physics and cosmology into sharp focus. It is not unusual when we examine a creationist argument we find that it might be good enough to fool the naive but it is not good enough to fool the knowledgeable.
First of all I will describe the underlying science and the experiment to put things in context and then I will review Humphreys’s article paragraph by paragraph.
The Underlying Science
The Humphreys article comments on popular press reports of a scientific paper published in Nature (2).
In the propagation of a polychromatic light pulse there are two velocities which are relevant. The first of these is the phase velocity of a monochromatic component, which is the propagation velocity of, say an electrical field maximum in the electromagnetic field. Where the phase velocity of different wavelengths is the same (ie in a medium of zero dispersion, such as light in vacuum) then the group velocity of the polychromatic pulse (the velocity with which the pulse propagates) equals the phase velocity. For light in a vacuum, this is equal to c, the speed of light in a vacuum.
In a dispersive medium (one in which the phase velocity is a function of wavelength) the phase velocity and the group velocity of a pulse are not equal. In all transparent passive media of normal dispersion, the refractive index increases with increasing frequency. The classical interference of waves results in a group velocity that is less than the slowest phase velocity of the highest frequency component. The appropriate formula is:
Where Vg is the group velocity, Vp is the phase velocity, k is the wavenumber and n is the refractive index.
Now in an anomalously dispersive medium, where the refractive index decreases with frequency, not only can the group velocity exceed the phase velocity, it can be negative. This counterintuitive result means that the pulse is observed to propagate in a direction opposite to the propagation of energy and information. This happens if k/n*dn/dk is negative. There is one very important thing to notice here – all of these results arise according to classical wave theory. There is nothing here to raise an eyebrow of any physicist.
(If one wants to understand the physical basis of a pulse propagating in the opposite direction to the phase velocity and the flow of energy, first consider that a narrow polychromatic pulse can be decomposed into a superimposed series of much longer monochromatic wave packets in the Fourier domain. In fact the narrow polychromatic pulse is necessarily composed of long monochromatic wave packets. For a pulse propagating in an anomalously dispersive medium the superposition of the long monochromatic waves produces a short polychromatic pulse propagating in a direction opposite to the direction of energy and phase propagation. The exact composition of these wave packets can be determined by a Fourier analysis of the polychromatic pulse)
So it is clear from classical wave theory that an anomalously dispersive medium can produce a group velocity that is greater than the phase velocity – in fact, such a medium can produce a negative group velocity.
The fly in the empirical ointment is that for all known transparent media operating in a linear domain, the dispersion is not anomalous resulting in a group velocity less than the phase velocity and a perfectly intuitive result. Wang et al addressed themselves to creating a medium with anomalous dispersion.
In order to do so they use non-linear optical phenomena. The medium is a gas of caesium atoms at 30 degrees centigrade, constrained in a 6cm long Pyrex glass cell.. A static magnetic field is applied to the caesium atoms to align the dipole moments. The atoms have three states : an excited state |0> and two ground states |1> and |2>. All atoms are prepared to be in the ground state |1>by optical pumping (two beams are used: one pumping the D2 transition to empty the 6S1/2F = 3 hyperfine ground state and the other to pump via the 6P1/2 hyperfine excited state to the |1> ground state. So the majority of atoms in the cell are prepared into the |F = 4,M = -4> |1> ground state. They then expose the medium to two strong Raman pump beams separated by 2.7Mhz. The Raman beams pump ground state |1> up to excited state |0> but are detuned from the |1> to |0> transition by an average large D 0 and from each other by 2D (2.7Mhz in the actual experiment). A further light beam, which is the probe field and which can be pulsed or CW stimulates the relaxation of excited state |0> into ground state |2> |F = 4, m = -2> which serves as the Raman transition final state. There are two obvious frequencies in which the gain in the probe field is maximised: ie when the probe field is resonant with the Raman transitions caused by the two pump fields.
As the probe field is detuned it therefore passes through two regions of high gain. The authors theoretically show that the optical susceptibility of the probe field is such that it experiences an anomalous dispersion region between the two gain lines.
The practical set-up includes the caesium gas contained in a cell, the two ground state pumping beams, the two Raman pumping beams and a probe beam. A beamsplitter is introduced to the probe beam on the source side of the cell so that a Detector D1 can sample the pulse before it propagates through the cell. The sample beam is detected after the cell by detector D2.
The authors measure a pulse in D2 which is advanced by 62ns when the Raman probes are tuned on resonance compared with when they are off resonance but is otherwise identical in both states. Since the phase velocity transition time of the caesium cell is only 0.2ns it seems like the peak of the pulse leaves the cell long before the it enters the cell. The group velocity index is – 310 ± 5 (– 62 / 0.2) At first sight it seems that causality is violated. However, a knowledge of classical wave theory immediately indicates the monochromatic components of the short polychromatic pulse are much longer that 62ns and that the negative group velocity index is an expected consequence of classical wave theory. Neither information nor energy propagates faster than the phase velocity and certainly not faster than c, the speed of light in vacuum.
where n is the frequency, n(n ) is the refractive index at frequency n and c the speed of light in vacuum.
So what have we got here?: it is a very cleverly constructed medium with anomalous dispersion, consisting of a gas of caesium atoms in a 6 cm long cell, with four separate pump beams and a probe beam creating an empirically challenging but classically predicted wave phenomenon.
There are some simple introductions to the wave propagation principles at:
Let's turn now to Humphreys article: see below. I have highlighted Humphreys's words in blue.
'The Speed of Light:
What do Recent Experimental Results Reveal?
by Physicist, Dr. Russ Humphreys
I've been pondering this experiment ever since I saw the technical article [L. J. Wang, A. Kuzmich, and A. Dogariu, "Gain-assisted superluminal light propagation", Nature, Vol. 406, pp. 277-279] on the Nature Web site on July 19th. The most puzzling thing to me is how the authors appear to deny the obvious implications of their data. They imply that their results do not suggest that information could be transmitted faster than the speed of light in vacuum, and yet the nearly-raw data in their figure 4 says just the opposite.'
As we can see from the analysis I have given above in the Underlying Science, the 'nearly-raw data' does not say any such thing: a knowledge of classical wave theory indicates that neither energy nor information is being transmitted faster than c. Humphreys is either mistakenly confusing group and phase velocity or he is deliberately obfuscating between the two.
The figure shows two pulses, A and B, several microseconds wide and of nearly the same shape. Pulse B has traversed their cesium vapor cell, and pulse A has traversed the same distance (6 cm) in a
vacuum, requiring the normal 0.2 nanoseconds to do so. Essentially every point of pulse B has arrived at the detector about 60 nanoseconds "earlier" than the corresponding point of pulse A. The completeness of the advance of pulse B implies we could indeed use it to transmit information faster than the speed of light in a vacuum.
Humphreys is confused here: no information is propagating faster than c, as the speed at which information is carried by a light pulse should be defined as the frontal velocity of a step function signal. In an anomalously dispersive medium the group velocity is given by the superposition of all the frequency components making up a polychromatic pulse and information cannot travel faster than the phase velocity of the slowest component. Furthermore, complex information requires the propagation of more than one pulse. The arrival of a train of pulses cannot occur faster than c/n.
He is also confused about the way the experiment is done. Has he actually read the paper? Both pulses have traversed the caesium vapour cell. In one case (A) the Raman pump and probe beams are detuned from the caesium D2 transitions by 2.5GHz thus providing a reference propagation through a normal medium. In the second case (B) the Raman beams are tuned to the transition thus creating the anomalous dispersion condition. Where he got the idea that A propagates through a vacuum, it is impossible to say.
The astute reader may have noticed something even more weird about the above numbers: pulse B seems to have gotten way ahead of itself! A close-up view in the article (Figure 4, inset) shows that the leading edge of pulse B emerges from the cesium cell about 60 nanoseconds before the leading edge of the pulse allegedly causing it enters the cell. The newspapers actually got that point right. This raises the possibility of transmitting information "backwards" in time. That would be astonishing! It would bring into question the standard interpretation of scientific causality; a child might send a warning decades into the past, telling his father not to beget any children. What happens if the father heeds the warning?
Humphreys speculates wildly. His fundamental error (and it is pretty fundamental) is to equate the velocity of information transmission with the group velocity. The length of monochromatic wave packets contributing to the polychromatic pulse exceeds the 62ns of the separation between pulses A and B (this is also a natural consequence of classical wave theory).
He is wrong about the information transmission and this speculation about children sending information decades back is childish nonsense unjustified by the theoretical considerations or the experimental data.
Despite the astonishing implications, the authors write: "The observed superluminal light pulse propagation is not at odds with causality or special relativity..." because, they say, the equation on which they based the subsequent analysis "itself is based on the causality requirements of electromagnetic responses." But if the data itself says clearly that pulse B arrived before pulse A, then the theoretical basis of subsequent analysis is irrelevant.
Humphreys misunderstands what Wang et al are saying in the paper.
They said: 'The observed superluminal light pulse propagation is not at odds with causality, being a direct consequence of classical interference between its different frequency components in an anomalous dispersion region'
They also said:
'We emphasize that the observed superluminal light propagation is a result of the wave nature of light. It can be understood by the classical theory of wave propagation in an anomalous dispersion region where interference between different frequency components produces this rather counterintuitive effect.'
These are plain statements that any physicist should understand.
They also say:, 'the observed superluminal light pulse propagation is not at odds with causality or special relativity. In fact, the very existence of the lossless anomalous dispersion region is a result of the Kramers-Kronig relation which itself is based on the causality requirements of electromagnetic responses'. So again Humphreys misunderstands or misrepresents the situation. It is not the 'theoretical basis of subsequent analysis' (as Humphreys would have it) on which the experiment depends, but the very existence of the anomalous dispersion region in the caesium which relies on a causal relationship between the electromagnetic fields in the cell and the real and imaginary parts of the dielectric constant. [The words Humphreys removed are restored in italics]
The authors' statement that ''the observed superluminal light pulse propagation is not at odds with causality or special relativity' is precisely correct according to classical wave theory and they point out that the ability to create an anomalous dispersion relies on the conventional laws of causality. Causality is not compromised in this experiment.
Humphreys: I wonder if the authors put in these mollifying words to help the paper get past the reviewers.
No, the authors did not put in these words to mollify. Humphreys insults the intelligence of the authors, the reviewers and the editors of Nature. The authors put in these words because they are correct according to classical wave theory and because any competent person considering them, including the peer reviewers would understand them.
Humphreys: But the discordance between the authors' data and their mollifying words has produced confusion in the media.
There is no discordance between the author's data and anything that they say. If Humphreys thinks so, it is either because he does not understand the work or because he is deliberately misrepresenting it. Meanwhile, it is not the purpose of, or a requirement for research papers in Nature to clarify matters for journalists (or self-styled Physicists and Cosmologists for that matter). I don't expect many journalists to really understand the implications of this work.
Humphreys continues: One critic of evolution and the media wryly commented to me that the news reports combine two discordant themes: "This is new and unexpected,"; and "This is nothing new and there is nothing to worry about."
And why should we care about what any particular critic thinks unless he is qualified to actually understand the significance of the work? The work is theoretically unremarkable and empirically very clever.
He says: One experimental stumbling-point is that there might have been an undetected low-level precursor pulse which entered the cell before the main pulse, and that the emerging pulse was caused by the precursor. That possibility must be examined carefully, but I would have thought the Nature reviewers already investigated that angle. But a problem with the "precursor" view is how it (or a tiny piece of the leading edge) could tell the cell how to reconstruct an almost-identical copy of the main pulse.
Humphreys demonstrates here his inadvertent or deliberate ignorance of the physics. Not only is there a precursor, but the constituent wave packets span a longer period than 62nsec. The actual pulse width is 3.7 msec FWHM ( 50 times longer than the pulse advancement) and the observed pulse advancement is a consequence of interference effects caused by the anomalous dispersion of the medium.
Humphreys: I've asked some of my physicist colleagues here at Sandia National Laboratories, ones who are experimentalists and laser-optics experts, to comment on the paper. Their first response was incredulity, but it's hard to imagine that the reviewers for Nature wouldn't be just as expert and just as incredulous. The paper makes an extraordinary claim which requires extraordinary proof. We only have the first stage of such proof so far.
Humphreys at this stage resorts to second hand reports of what his colleagues think. Perhaps he feels he needs the credibility of the Sandia name. I have failed to see any Sandia National Laboratories scientists (other than Humphreys) come out in public with such statements. Anyone who has any expertise in this matter (including Humphreys's colleagues at Sandia) has no need to be incredulous. Certainly I am not, nor does it seem that the authors, peer reviewers and editors of the paper are. No-one who understands classical wave theory and the very clever empirical arrangement need be incredulous. Humphreys either does not understand these things or has set out to deliberately mislead the naive. I understand that Humphreys left Sandia to become a full-time faculty member of the Institute for Creation Research in October 2001, a little more than a year after he published this article on AiG's website. (It is interesting to note that his profile on AiG does not make it clear that he is no longer affiliated to Sandia National Laboratories).
Humphreys adds: Another odd thing is how fast the paper went through peer review. The facts of the article are so counter-establishment that I would have thought the pro-establishment Nature would have taken a year or so to approve it.
Well actually, the results are far from counter-establishment. The theoretical basis for a group velocity exceeding c has been well established for decades and can be found in the text books that some of us learn optics and electrodynamics from: Born and Wolf, Principles of Optics; and Landau and Lifshitz, Electrodynamics of Continuous Media. The experimental creation of an anomalous dispersion medium is clever but not revolutionary.
For example, one paper supporting very rapid reversals of the earth's magnetic field took several years to appear in the journal. It amazes me that it only took a little over a month for this one.
It was fast because it is uncontroversial. Humphreys is seeking ignorantly or deliberately to create controversy where none exists.
He continues: So either the experiment marks a new epoch in physics (communication faster than light and backwards in time), or it is a colossal blunder by the experimenters, their reviewers, and Nature.
This is a false dilemma. It is extremely interesting work but it is neither a new epoch in physics ( the group velocity is not the speed of information propagation) nor is it colossal blunder. The authors are absolutely correct in their interpretation of what they are doing. It belongs in Nature because it is the first time that an anomalous dispersive medium has been created with these properties.
He turns to accusations of incompetence (that's rich coming from him after this display): It's pretty hard to make a 60-nanosecond mistake in today's laboratories. Future experiments by other researchers may resolve the possibility of a mistake.
There is no mistake
Let's look at the other possibility, that Wang, et al., are right. What would happen to physics, then?
Humphreys now again enters hyperspeculation mode, not understanding or not choosing to understand that the group velocity is not the velocity of information propagation.
First, let me dispose of a wrong impression: relativity by itself does not exclude the possibility of things traveling faster than light. Just look up "tachyons" in the physics literature of a decade ago to convince yourself of that. Tachyons were hypothetical particles which obeyed relativity by "always" traveling faster than light, and never slowing down to that speed.
The idea of things traveling backwards in time is not forbidden, either. Recently I was looking at an old paper by famous physicist Richard Feynmann in which he had proposed that antiparticles positrons, antiprotons) are just ordinary particles (electrons, protons) traveling backwards in time. Everyone took him seriously; they even gave him a Nobel prize for the theory. Today, fifty years later, it is still a possibility.
The real problem is causality. What happens if the kid's father decides not to beget the kid? It seems to me that we could wind up seeking a broader interpretation of causality. Science-fiction authors have speculated about this for years. For example, two short stories by Robert Heinlein, "By his Bootstraps" and "Elsewhen," present two very different possibilities about the nature of time, while still allowing time-travel and resolving the causality problem.
Humphreys big problem is that this experiment is no challenge to causality. He either hasn't understood it or he is deliberately misrepresenting it. (Interesting that he misspells Richard Feynman's name).
He preaches: More significantly, for millennia the Bible has been transmitting detailed information to us about the future. I haven't noticed the world collapsing into non-causal chaos quite yet!
Nor have Wang, Kuzmich, and Dogariu. The relevance of the bible to this topic is quite impossible to discern.
More preaching: So here's an opportunity for physicists, philosophers, and theologians to shamelessly speculate about a possibly real phenomenon. But don't go investing in time machines just yet. I think the odds are high that a mundane, dull, non-exciting explanation will emerge.
That explanation is already contained in the Nature paper. The phenomenon is caused by classical interference of waves in a medium with anomalous dispersion. If Humphreys was really a competent cosmologist and physicist he would understand this.
His parting shot: But there is always the possibility that God has slipped another surprise for us into the world He created. That's why physics is fun for me.
Physics might be fun for Humphreys but in this article, he seems to be either ignorant about the basics of it or deliberately misrepresenting it.
The Humphreys article is supported by this absurd comment from the 'editor'
Editor's comment: 'Whatever the outcome, it is a handy reminder that we don't know as much as we think we know about such things as how light really behaves or propagates in deep space, the nature of elementary particles, and so on, which is all very relevant to issues, such as starlight travel-time, radiometric "dating" processes, etc. The dogmatic certainty with which many researchers and popularizers write about such things as this, and about an alleged 'big bang', for instance, is not warranted by the data. There are a number of physicists at the 'cutting edge' of such things who prefer a more humble approach.'
There is of course no relevance or connection between the editor's preaching above and the Wang et al paper on superluminal propagation.
As we can see here Humphreys is either ignorant of the physics or he deliberately misrepresents it. He certainly does not seem to have read the paper very carefully before fulminating in public.
We have seen that the creationists' physics champion seems to have feet of clay - or he is deliberately misleading the naive readers of Answers in Genesis. There appears
to be no other possibility. This is the man who claims to have an alternative
cosmology to compete with more standard models. Is that likely in someone
who apparently cannot understand electromagnetic wave propagation at the graduate
Revised 21st February 2003
Postscript - 29th November 2003
When the original article above was published on the precursor to this website, I sent an e-mail to Russ Humphreys at Answers in Genesis to let him know of its existence and to give him the opportunity to reply. On 29th January 2003, he dismissed my critique as follows:
This is about ancient news. That is, Mr. MacAndrew has spent a lot of time making a critique of a quick comment of mine about a flash in the pan that occurred in the science journals several years ago. I haven't seen anything in the journals on the subject since that time. Please tell Mr. MacAndrews that nobody cares anymore. --- Russ
I replied on 30th January 2003:
Actually it is just 2.5 years ago. And it's still relevant, because it's still on your website and it's still being used by creationists to attempt to demonstrate that the speed of light and the transmission of information can be anything you like in support of a Young Universe. Furthermore, obvious misunderstanding of a subject is always relevant if the perpetrator claims to be knowledgeable in it.
Interesting that what Mr Humphreys referred to in August 2000 as 'either the experiment marks a new epoch in physics (communication faster than light and backwards in time), or it is a colossal blunder by the experimenters, their reviewers, and Nature.' is now a 'flash in the pan'. Apparently physics is not as much fun for Mr Humphreys as he claimed in the article.
I fully sympathise with Mr Humphreys desire to sweep this under the carpet. I'd be as embarrassed about it as he is if I'd been caught out like this. The article might not be news anymore but Mr Humphreys's fundamental physics errors will always be relevant as long as he claims to be a knowledgeable physicist and cosmologist.
I received no reply to this last message.
Well, Russ Humphreys has been proved wrong again. His judgment about what is and isn't important physics is actually very poor. Far from being 'a flash in the pan', research has continued in this topic. On 16th October 2003, Stenner et al published an important paper in Nature (5), describing an experiment to measure the speed of information in a medium where the group velocity of light pulses propagates superluminally.
Stenner et al measured the time to detect information propagated by a pulse which traverses either a vacuum or a region of anomalous dispersion. The leading edge of a Gaussian pulse is advanced by 27.4 nsec in a similar way to the experiment described above (although Stenner et al use a cell containing a vapour of potassium atoms rather than caesium atoms as used by Wang et al in the original work). However when the amplitude of the Gaussian function is switched either to a high amplitude (corresponding to a '1' bit), or a low amplitude (corresponding to a zero bit) near the peak of the Gaussian, the zero or one bit information is detected in the superluminal case slightly later than the case where the pulse traverses a vacuum, even though the group velocity of the pulse in the superluminal case vastly exceeds the speed of light. In other words, strict causality as required by special relativity is preserved.
The conclusion is therefore that information does not propagate faster than the speed of light even in cases where the group velocity is higher than c, the speed of light.
So Russ Humphreys might think that no-one cares anymore, but that is probably because he made so many fundamental errors in his article. It's a pretty damning piece of evidence that someone who claims to be a knowledgeable physicist should be so ignorant about hot topics in modern physics. He says that he hasn't seen anything on the topic since 2000. In fact papers on superluminal propagation and causality continued and continue to be published at a high rate (6) - (11).
Mr Humpheys claims to be a competent physicist and cosmologist. He is a talisman for young earth creationists who hang on to his 'alternative cosmology'. We have seen here, however that he is incompetent in a topic that he himself chose to address. Why should we take his 'alternative cosmology' any more seriously than we take these error-ridden attempts to comment on superluminal pulse propagation?
2 Wang, Kuzmich, and Dogariu, Gain-assisted superluminal light propagation, Nature 406, 277-279 (2000)
(The Nature paper is fully referenced to relevant prior work and the supporting science)
5 Stenner, Gauthier and Nelfeld, The speed of information on a 'fast-light' optical medium, Nature 425, 695 - 698 (2003)
6 Nimitz and Gaibel, Basics of superluminal signals, Ann Phys 11, 163 - 171 (2002)
7 Akulshin et al, Negative group velocity of a light pulse in cesium vapor, Quantum Electron 32, 567 - 569 (2002)
8 Bigelow et al, Superluminal and slow light propagation in a room temperature solid, Science 301, 200 -202 (2003)
9 Macke and Ségard, Propagation of light pulses at a negative group velocity, Eur Phys J D 23, 125 - 141 (2003)
10 Wynne, Causality and the nature of information, Opt Commun 209, 85 - 100 (2002)
11 Solli, Chiao and Hickmann, Superluminal effects and negative group delays in electronics and their applications, Phys Rev E 66, 056601 (2002)
12 Kazmich et al, Signal velocity, causality and quatum noise in superluminal light pulse propagation, Phys Rev Lett86, 3925 - 3929 (2001)