Not long ago, a new preprint on the fine structure constant got a bunch of press, nicely summed up by the Knight Science Journalism Tracker last week. I meant to say something about this last week, but what with it being the first week of classes and all, I didn’t find the time.
I still think it’s worth writing about, though, so after a reproduction of the key figure, we’ll have the usual Q&A-format explanation of why I don’t quite trust this result:
So what’s this all about? The preprint in question is the latest in a series of attempts to measure possible changes in the fine structure constant by looking at the spectra of distant galaxies. Not only do they seem to see a change in the constant, the change seems to be different in different parts of the sky.
Back up a minute. A constant is changing? Well, a dimensionless ratio of fundamental constants, though some people would say that those are the only constants that matter. We’ve talked about this sort of thing before, but the short version is that some theories predict that the ratio of the electron charge squared divided by Planck’s constant and the speed of light could change over the history of the universe. This paper claims that they have not only measured the change by looking at light from distant quasars, whose spectra should depend on the value of the constant billions of years ago, but see it changing differently in the northern sky than the southern.
Isn’t that awfully strange? If it’s a real effect, it would be really weird. The simplest model of the universe would have the constant really being constant, and the first correction you would expect to see to that would be for it to change by the same amount everywhere. Having it change by different amounts in different places would be bizarre.
So, how do they measure this? The paper is an extension of an earlier method developed by this group, that is, shall we say, not without controversy. A different group using a similar technique got different results, prompting a re-analysis by the original group claiming the second paper was all wrong.
Isn’t this the reason why people like Cameron Neylon argue for open data and public access to analysis codes and that sort of thing? Yes, but that’s a different issue.
OK, so what is this method? They look at the light coming from a distant quasar, which passes through lots of different gas clouds along the way. Because the red shift of the light from the quasar depends on the distance it is away from us, each of these gas clouds absorbs light at very slightly different frequencies.
They look for absorption signals from two types of ions, magnesium and iron, and compare the positions of those lines in the spectra from specific gas clouds. If the shift of the lines in question is bigger or smaller than that associated with the expansion of the universe, they attribute that to a change in the fine structure constant, which plays a role in determining the energy states of the ions in question.
Why those two ions? I’m not sure why magnesium and iron specifically– probably it has to do with them having transitions at convenient wavelengths– but as a general matter, they compare two different species because their spectra are affected by changes in the fine structure constants in different ways. the relevant lines in magnesium hardly shift at all, while the iron lines are affected in a much larger way. That means they can use the magnesium lines as a marker to get the red shift associated with the position of the cloud in the universe, and see if the iron lines are shifted by more or less than that amount.
So they see different values in different places? Right. Their initial data were all taken with a single telescope (though by different groups of observers), and looked at sources primarily in the southern sky. Now, they’ve added data from a second telescope, which primarily covers the northern sky. The typical shift that they see in the northern sky is in a different direction than the typical shift they see in the southern sky. They suggest that this might indicate a spatial variation in the fine structure constant.
But you don’t believe it? It’s an extraordinary claim, and that demands extraordinary proof. And there’s one very simple reason why I don’t really trust this result, demonstrated nicely in the figure above.
OK, what’s the figure? The figure is a plot of the sources they looked at on the sky, in the funny projection that astronomers use, with the shape of the symbols indicating which telescope was used for the observation, and the color of the points indicating the sign and strength of the shift (more pink is a bigger shift in one direction, more blue is a bigger shift in the other direction. As you can see, points in the north are mostly circles, and mostly pinkdish, while points in the south are mostly squares, and mostly bluish.
Yes, and? There are also a bunch of triangles on the plot, which are sources they looked at with both telescopes. These are mostly in the boundary region between the two, and the striking thing about them is that they’re nearly all black, indicating no change. Even the ones that are well away from the boundary between blue and pink areas are black, including one that’s right on top of a big pink square.
So? They’re in the boundary region. It just seems awfully convenient that the difference they see is so well correlated with the specific telescope they used. There’s no particular reason why the changing constant should align with the location of observatories on our fairly insignificant little planet.
They make a passing reference to this, but try to brush it aside by saying “To explain our results in terms of systematics will require at least 2 different and finely tuned effects.” I don’t find that terribly convincing, though– I think it’s much easier to believe that one telescope’s data set was coming out slightly high, and the other’s slightly low, in a way that they more or less cancel each other out than that this is really a spatial variation in a fundamental constant.
So, how do we sort this out? Well, they’ve got the right approach in the next sentences of their paper:
Future similar measurements targeting the apparent poleand anti-pole directions will maximise detection sensitivity, and further observations duplicated on 2 independent telescopes will better constrain systematics. Above all, an independent technique is required to check these results.
We need to see more observations from different telescopes, and see if the effect appears the same. And we also need somebody else to come up with another technique that can be used to look for changes in different parts of the sky. If other people using a different technique find the same sort of changes and the same sort of distribution of changes, then it needs to be taken seriously. Until then, I remain skeptical.
I also find this evidence much to weak to prove anything, but the idea of a fine structure constant changing from place to place doesn’t seem all that strange to me. After all it’s definition involves electric charge which is sensitive to vacuum polarization.
If you accept that vacuum fluctuations may in fact not be random but may depend weakly on nearby matter and fields then it would follow that local values of charge and alpha may differ a bit from place to place.
After all alpha also depends on effective energy of the interaction so we already know it’s not strictly constant.
I haven’t thought about this much, but I’m also not sure why time variation would be more natural than spatial variation of the fine structure constant. In particular, it seems to me that this statement depends on frame of reference, and why would whichever effect that generates the variation align precisely with our (current) reference frame’s notion of time?
The simplest model of the universe would have the constant really being constant, and the first correction you would expect to see to that would be for it to change by the same amount everywhere. Having it change by different amounts in different places would be bizarre.
I don’t agree; if it changes over time, then it’s probably the expectation value of some scalar field, and relativity demands that it can also vary over space.
In particular, it seems to me that this statement depends on frame of reference, and why would whichever effect that generates the variation align precisely with our (current) reference frame’s notion of time?
In a Big Bang cosmology there is a particular frame of reference in which the dipole term in the cosmic microwave background vanishes (there are higher order multipole terms). We can detect our own motion with respect to this frame because it creates a blueshift of the CMB in the direction we are moving and a redshift in the opposite direction. It makes sense to me that you might have a purely temporal variation in the fine structure constant as viewed in this preferred frame. Our actual motion with respect to this frame is sufficiently slow that any resulting spatial variation in our frame might be well within the error bars on the measurement. Not saying that this is correct, only that it is plausible given what we know.
Chad is right to be skeptical of the result. Does our velocity with respect to the spherically symmetric frame correspond with the respective fields of view of the two telescopes? If not, then I would look into systematics with the two telescopes. If yes, then the result is plausible but still needs checks with other telescopes before it could be confirmed.
I agree with Chad. Skepticism is warranted. Since electronic transition energies scale like alpha^2, and I believe the 21 cm hyperfine transition energy scales like alpha^3, any real effect would cause a very noticeable offset between 21 cm line and optical line redshifts that has never been seen. There are other constants involved in the 21 cm line but it would take a conspiracy to avoid a detectable effect. See http://www.astro.ucla.edu/~wright/old_new_cosmo.html#05Mar07
Thanks Eric. Yeah, this sounds plausible, but I am missing a few details. I’m curious enough that I might go chat with one of the local experts.
You may be skeptical, Chad, but what we really want to know is if your dog is skeptical….
Alpha is not a fundamental constant, alpha is a ratio: (in cgs) alpha = (e^2)/(h-bar)c. However, this is not alpha either, hence exploitable loopholes,
alpha = (e^2)/4(pi)(e_0)(h-bar)c
or
alpha = (e^2)c(mu_0)/2h
The cgs statcoulomb is defined with [4(pi)(e_0)] = 1 or with a like manipulation of mu_0, the electric permittivity or magnetic permeability of free space respectively. Pi, electrons, lightspeed, and Planck’s constant are robust values.
e_0 and mu_0 can be altered at will, e.g., inside a Casimir gap as the Scharnhorst effect (Rabi vacuum oscillation, electron anomalous g-factor, etc.). Alpha may be a weak function of what, how much, and how space is filled with matter and fields. SWAG time! Test alpha as Casimatter – a lump of stuff that is *only* Casimir etalons.
Casimir force varies as 1/d^4, “d” being reflector spacing. Yer gonna need a very small gap for measurable high order anomalies.
The smallest fabricatable solid transparent Casimir gap is optical half-pathlength ~60 nm created by 70 nm-thick aluminum reflectors spaced with 37 nm of 60:40 MgF2:LiF alloy, weighted refractive index (at 120 nm) 1.628. The alloy matches aluminum’s linear coefficient of thermal expansion, 23.1 ppm/K. Smaller gaps (shorter wavelengths) do not have efficient reflectors (95+% reflectance for oxide-free Al at 120 nm) or transparent spacers (95+% transmission for the fluoride alloy).
Imagine a flat torus in hard vacuum spinning just above a cluster of alternate magnetron sputtering sectors of Al and 60:40 MgF2:LiF alloy. A continuous flat spiral bifilar deposition results: Al, optical alloy, Al, optical alloy… Cut out a section to have Casimatter, stuff that is nothing but a stack of Casimir etalons. Work the material densities to see that 37 wt-% is vacuum zero point fluctuation-depleted fluoride alloy.
Play with that (e.g., inject F-centers and look for electronic anomalies). Physics demands chemistry is derivable and therefore irrelevant. Ask a physicist for an aspirin. Does he derive it or open a bottle?
Yes, it’s these two sentences that make “Changing Fine-Structure Constant” worthy of my second read. “Future similar measurements targeting the apparent pole and anti-pole directions will maximise detection sensitivity, and further observations duplicated on 2 independent telescopes will better constrain systematics. Above all, an independent technique is required to check these results.”
That last sentence is what research, no matter what field, leads to: further inquiry into other possibilities. Thanks for your post, John. A good exercise for my “retired” brain.
While I’m not gonna hop on the bandwagon and believe in a changing α, I’d still like to ask if it isn’t a bit too skeptical to dismiss an effect just because it’s second order? I’ve had enough grouptheory/spectroscopy to remember that first order effects may well cancel out on symmetry grounds. Could that not be the case here as well? The s component being somehow disallowed so only the p bits show up. Or am I completely missing the point and imposing a non-relativistic viewpoint?
If you look at the linear regression plot of shifts versus polar angle you’ll see (wearing my statistics hat now :>) ) heteroscedascity (sp?), i.e. the deviations of the points from the straight line don’t seem to follow a random pattern, but are clustered on one side of the line for the southern hemisphere and on the other for the northern, with extreme deviations for the two polar points. That to me speaks of some sort of systemic error.
Well, now I read the Mr. Orzel’s objection and it’s all actually based on this single sentence
/* …It just seems awfully convenient that the difference they see is so well correlated with the specific telescope they used. There’s no particular reason why the changing constant should align with the location of observatories on our fairly insignificant little planet. ..*/
Actually, such reason may exist quite easily if we realize, the center of anisotropy is located at the north hemisphere, which isn’t observable with telescopes at southern hemisphere – and vice versa. So that it’s nothing strange, when telescopes at the north hemisphere would observe positive shift, whereas the telescopes at southern hemisphere would observe negative shift. Mystery explained.
It’s evident, the whole problem is in the point, Mr. Orzel has no idea about experimental background of astronomical surveys – or maybe he considers Earth flat or whatever else.
My problem is rather the following one:
Well, if fine structure constant is characterizing the strength of the electromagnetic interaction, then my question is, whether it should change at GUT scale, when all interactions converge together.
If yes, then the question is, why fine structure constant shouldn’t change at the remote areas of Universe, when the Universe was close to GUT scale. I hope, the question defined in such way is sufficiently clear for everyone.
I have now updated the Fine-Structue constant wikipedia page quoting yourself. If you want to change anything or make it more explicitly, please check http://en.wikipedia.org/wiki/Fine-structure_constant#Is_the_fine_structure_constant_actually_constant.3F
Thanks, and long life to the constant-constant!
Spatial variation of the fine-structure constant?
J. K. Webb et al., arXiv: 1008.3907v1 presented possible evidence of a spatial variation of the fine-structure constant, where the axis of the dipole points to R. A. = 17.3h, dec. = -61°.
Such a spatial variation, if confirmed, might indicate an anisotropic universe. I would like to point out two earlier works which reported possible evidence of an anisotropic universe.
P. Birch, Nature 298 (1982) 451-454 presented possible evidence of a vorticity of the universe, where the axis of the dipole points to R. A. = 14h 55min, dec. = -35°.
Only a small part of the 3K dipole can be explained by the motion of the Sun around the Galactic centre and the gravitational infall of the Milky Way into the Virgo cluster of galaxies. A. Dressler, Nature 350 (1991) 391-397 suggested a motion of the Local Supercluster towards Galactic longitude l = 307° and Galactic latitude b = 9° (approximately R. A. = 13.5h, dec. = -45°). His claimed Great Attractor has never been detected. So it is possible that this so far unexplained part of the 3K dipole results not from Local Supercluster motion, but from an anisotropic universe.
The three directions listed above differ from one another. However, the error bars are large. Possibly the works of Birch, Dressler, and Webb et al. support an anisotropic universe.
Anyone who is interested in my early work on an anisotropic universe is invited to read my paper R. W. Kühne, Mod. Phys. Lett. A 12 (1997) 2473-2474 = arXiv: astro-ph/9708109. In it I argued that the alignment of the rotation axes of the galaxies of the Perseus-Pisces supercluster results from universal vorticity (Gödel cosmology).
Anyone who is interested in my early work on a time-variation of the fine-structure constant is invited to read my paper R. W. Kühne, Mod. Phys. Lett. A 14 (1999) 1917-1922 = arXiv: astro-ph/9908356.
I like to call all of the constants and the endless coincidences around them that lead to life the “amazing set”
Are we assuming that there is only one workable set? If this problem is looked at from the point of view that constants don’t start out constant but settle into a stable set later then looking at this problem of oddities in the fine structure constant is a clue to how stable universes formed after the big bang. What if there were endless expressions at the first moment, but only the stable arrangements of constants survived into the next moment.
Here is a long shot…. Life stabilizes sets of constants. Energy and matter only universes don’t need a time constant or the fine structure constant to constrain or order them. Living things do. So, total chaos may reign over most of the multi-universes but constant derived fractal chaos reigns in universes with life.
Have fun
Chad, I think you need to correct this discussion since you neglected to mention one feature of this article – and even of the figure you show here – that refutes your skepticism.
I am talking about the blue, green and red blotches which represent the direction of the dipole, as estimated *independently* by the two telescopes (blue/green) and then in the joint data set.
If the result were purely a telescope-dependent systematic, you would not expect to see any significant dipole in either telescope’s data taken separately. You would expect a small dipole simply due to random fluctuations, but the directions would not be expected to agree between the two.
But in fact, the two independently estimated dipole directions do agree to within their errors, and are consistent with the dipole direction of the total data set (red blotch). That seems a quite improbable occurrence just due to random chance.
The other check which you don’t mention is that some systems are observed by both telescopes.
The recently published PRL article states: ‘The independent VLT and Keck samples give consistent dipole directions and amplitudes, as do high and low redshift samples. A search for systematics, using observations duplicated at both telescopes, reveals none so far which emulate this result.’
I don’t feel any need to correct this, because I still feel the same as I did a year ago. I’m pretty dubious about the process of fitting a whole-sky dipole pattern to a scattering of data points in only half the sky, and don’t find the coincidence of the alignment of the poles to be more disturbing than the notion that the poles of the universal fundamental constants would happen to line up fairly nicely with the poles of the Earth.
As for the sources observed by both telescopes, I dealt with that in the post above: the sources observed by both are necessarily in the boundary region between telescopes, and mostly show no significant shift. Which is exactly what you would expect from a systematic shift between telescopes: if one is systematically high and the other systematically low, the sources seen by both, which presumably use some sort of average of the two values, would show little to no shift.
Finally, in a less scientific and more sociological area, I am unhappy about the fact that the figure showing the individual sources is only a “supplemental” figure in the preprint, and does not appear in the actual paper (the full-sky map figure in the paper shows only the colored dipole blobs). In fact, there are basically no details about the individual sources in the paper, with all information about the individual measurements and the tests for systematic errors pushed off into a pair of Hopeful Future Citations (one “submitted” one “to be submitted”). Honestly, I have no idea how you could even referee the recently published paper, with what seems to me to be critical information entirely absent from the paper.
So, no, I don’t feel a need to correct my discussion. I think it’s fine the way it is.