## Monday, July 14, 2008

### Ammonia: Proton-electron mass ratio constant for 6 gigayears

In 2006, a Dutch team (Ubachs et al.) measured the Hydrogen spectrum emitted 12 billion years ago and concluded that the proton-electron mass ratio - currently close to 1836.15 - used to be 1.00002 times higher than it is today!

I have always believed that the experiment had to be wrong because the constants (not only the mass ratio but also other functions of the low-energy parameters of the Standard Model) have no good reason to evolve. Moreover, it is very unnatural for the change during 90% of the universe's lifetime to be 20 parts per million only but not zero.

Most likely, the moduli have been stabilized since the very early moments after the Big Bang. Assuming otherwise would lead to all kinds of bizarre predictions, including new long-range forces or new phase transitions that could spoil the Big Bang cosmology or gradually destroy life by modifications of the binding energies of DNA bases. I simply don't buy such an evolution and I don't think that such an evolution really helps to solve the cosmological constant problem or anything of the sort (moreover, quintessence seems highly disfavored observationally).

Many journalists (and physicists!) occasionally use the term "string theory" when they mention their speculations about evolving constants of Nature. But it is important to know that every single universe from the proverbial "landscape" of 10^{500} or so solutions predicts that the constants are fixed forever. In canonical string theory, these constants don't evolve.

That's true almost by definition because the number counts the stabilized vacua. There could also exist semi-realistic non-stabilized vacua where the constants evolve but they seem to have problems, they are not terribly well understood, and most of the "good" physicists (this adjective shouldn't be interpreted too dogmatically) don't believe that they are relevant. We could still be wrong but it seems misleading to use the term "string theory" to support the speculations about evolving constants because string theory, as understood today, works differently and much more rigidly.

Back to 2008

Now, USA TODAY informs about a new experiment by Christian Henkel et al. (Bonn, Germany) that looked at ammonia molecules 6 billion years into the past.

Do you remember the Feynman lectures in physics, namely the chapter where quantum two-level systems are discussed?

The ammonia molecule has a nitrogen atom right above the center of the triangle of hydrogen atoms. The energy eigenstates are the sum and/or the difference of the quantum states where the nitrogen is above and/or beneath the triangle. And very low frequency photons can be emitted in between these two states. The frequency turns out to sensitively depend on the proton-electron mass ratio.

See their original paper in Science from June 2008:
Strong limit on a variable proton-to-electron mass ratio from molecules in the distant universe (full preprint plus a press release)

During the last 6 billion years, the ratio has changed, relatively speaking, by less than 2 parts per million (ppm) at the 95% confidence level, much less than the alleged Dutch positive signal that was claimed to be 20 parts per million.

Whenever there are two experiments and one of them gets an agreement and the other one gets disagreement, I think that it is much more likely that the experiment showing a disagreement is incorrect because it is much easier to introduce an error (any error) that destroys the agreement than to keep all things pure and accurate so that the agreement is preserved. ;-) In other words, it is unlikely to get an accurate agreement by chance.

Of course, the Dutch team has looked into a more distant past (something could have, in principle, happened when the Universe was between 1 and 8 billions years of age) and it has studied a different molecule (that could, in principle, lead to different results than the ammonia molecule). But I find all these explanations unlikely. The most plausible explanation is that the Dutch experiment was simply wrong. It would be nice if someone could locate the mistake more accurately.

If you believe the new paper, it might be the best bound available on the market.