Cosmic discord

Article metrics

New data on the cosmic background radiation are making cosmologists revise their view of the Universe. The biggest surprise is that the data favour a larger amount of 'ordinary matter' than was thought.

The early Universe is a simple system composed of nearly uniform ordinary matter, radiation and dark matter. Cosmologists say that an early period of rapidly accelerating expansion (inflation) created small-scale fluctuations that grew into the large and complicated structures we see today (galaxies, stars and planets). Larger-scale fluctuations are also imprinted on the cosmic microwave background, or CMB — the faint radiation left over from the Big Bang. Recent measurements of temperature variation in the CMB, especially from the Boomerang and Maxima experiments1,2, reveal distinctive patterns in these fluctuations, which depend on the details of the shape and composition of the Universe.

In several papers appearing on the web3,4,5,7 and now in print8, cosmologists are starting to interpret these patterns through detailed statistical studies. They are finding amazing consistency with the expectations of inflation — a nearly 'flat' Universe, which can be described as a small piece of an enormous hypersphere — and with independent estimates of quantities such as the densities of dark matter and 'dark energy'. But there are also a few unexpected and possibly important discrepancies.

The sharpest and most interesting discrepancy is the estimate of the density of ordinary (baryonic) matter in the Universe: the new data say that the mean number of neutrons and protons per unit volume is greater than was thought. The baryon density connects with several wildly different observations of the Universe. The mean density of baryons when the Universe was 300,000 years old affects the properties of primordial acoustic waves, which directly influence observable fluctuations of the CMB measured by Boomerang and Maxima. At much earlier times, at a cosmic age of about one second to a few minutes, the baryon density is the one parameter that determines the production of light nuclei — especially the fractions of deuterium (D), helium (He) and lithium (Li), whose primordial abundances are estimated from observations of stars and gas more than a billion years later9,10. Finally, we can estimate the mean density of baryons simply by looking around today and counting them — most of them are found in hot, ionized plasma on the outskirts of galaxies11.

Recent improvements in these three estimates have produced remarkable concordance. But close examination of the best current data now shows a statistically significant discrepancy: the CMB data imply a higher baryon density than the abundance of light nuclei does, and a reasonable amount of waffling in other cosmological parameters seems unable to eliminate the problem. It seems that we are bound to learn something new — perhaps about the interpretation of one of the data sets or one of the other cosmological parameters, or perhaps about the standard cosmological model, which is used to weave together the different observations (Box 1).

The first map of the CMB, from 1992, revealed large-scale fluctuations (at a scale of 10° to 90° across the sky), but Boomerang and Maxima can measure much smaller-angle temperature fluctuations. They find the expected broad 'acoustic peak' in the power spectrum at an angular scale of around 1°, followed by smaller peaks at even smaller scales. The relative amplitudes and positions of these peaks are sensitive to the baryon density because the baryons contribute the main inertia in the primordial acoustic waves responsible for the pattern of peaks. The current data conflict with the predictions of the standard cosmological model because the smaller peaks, especially the second acoustic peak, are much weaker than expected. By varying the favoured values of the cosmological parameters within the constraints of the data to explore the range of viable universes, it seems that no simple statistical fix can resolve the discrepancy8.

Immediate suspicion falls on the baryon density. A higher baryon density increases the amplitude of the odd peaks (the first and third) relative to the even ones (the second and fourth). If the baryon density is made a factor of two higher than the value indicated by the abundance of light nuclei, the second acoustic peak would be lowered in line with the Boomerang and Maxima data. Such an adjustment seems reasonable, because considerable interpretation is required to derive the baryon density from abundance data9,10.

Nonetheless, the baryon density preferred by the CMB data predicts nuclear abundances that are unacceptable to anyone. The best fit to the Boomerang data4 using the favoured Hubble constant of H0 = 71 ± 8 (which gives the expansion rate of the Universe in km s−1 Mpc−1), estimates a baryon density, Ωb, of 7.4 ± 1 %. This translates to a baryon/photon ratio (in dimensionless units of 10−9), η−9 = 1.0 ± 0.15. For η−9 = 1, standard models of nucleosynthesis during the Big Bang yield the following abundances for the light elements relative to hydrogen: D/H = 1.12 × 10−5, 7Li/H = 1.0 × 10−9 and a He abundance of 25.2% by mass.

These estimates do not agree with the available observations. The predicted D/H is lower than direct measurements of abundances in quasar absorbers, the interstellar medium, and the jovian atmosphere — a paradox because the Big Bang is thought to be the unique cosmic source for deuterium and subsequent evolution destroys it. The He abundance by mass is only about 1–2% above the observational estimates in metal-poor dwarf galaxies, but this is still thought to exceed the systematic measurement error. Current estimates of Li from metal-poor stars show a 'plateau' at Li/H 10−10 (ref. 12), which may be close to the primordial abundance. Although it might be possible to uniformly destroy 90% of the 7Li these stars started with, there are good arguments against this option — including the presence of 6Li, which is far more fragile.

There are other ways to remove the baryon-density discrepancy, but none of them is compelling. There may, for example, be a new source of reionization at very high redshift that creates a thicker electron-scattering photosphere through which we view the fluctuations. There might be a 'tilt' in the primordial spectrum that changes the initial angular dependence (although this would spoil the beautiful agreement of intermediate-scale fluctuations with both larger-scale fluctuations and galaxy clustering). Or other sources of dissipation could be introduced that would damp acoustic waves more quickly than standard theory predicts. Some of these possibilities can be ruled out by examining the smaller acoustic peaks. The tilt and damping options lead to all the smaller peaks being suppressed, whereas a higher baryon density predicts a return to a higher value for the third peak, as well as a shift in the location of the peaks owing to the difference in sound speed.

It is surprising that our various measures of baryon density agree so well. After all, we are comparing evidence from processes that operate at cosmic ages and scales differing by more than ten orders of magnitude, and find values for the baryon density that agree to within a factor of two. Nonetheless, exploring this discrepancy might lead to something really new — perhaps a simple reinterpretation of data on abundances, or perhaps a new ingredient not yet included in the standard cosmological model.


  1. 1

    De Bernardis, P. et al. Nature 404, 955– 959 (2000).

  2. 2

    Hanany, S. et al.

  3. 3

    Lange, A. E. et al.

  4. 4

    Balbi, A. et al.

  5. 5

    Hu, W., Fukugita, M., Zaldarriaga, M. & Tegmark, M.

  6. 6

    Lesgourges, J. & Peloso, M.

  7. 7

    White, M., Scott, D. & Pierpaoli, E.

  8. 8

    Tegmark, M. & Zaldarriaga, M. Phys. Rev. Lett. 85, 2240–2243 (2000).

  9. 9

    Nollett, K. M. & Burles, S. Phys. Rev. D 61, 123505 (2000).

  10. 10

    Burles, S., Nollett, K. M. & Turner, M. S.

  11. 11

    Fukugita, M., Hogan, C. J. & Peebles, P. J. E. Astrophys. J. 508, 518– 530 (1998).

  12. 12

    Ryan, S. G. et al. Astrophys. J. 523, 654– 677 (1999).

Download references

Author information

Correspondence to Craig J. Hogan.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.