Measuring the expansion of the Universe is like checking the time on two perfect clocks: one set by the ancient light of the cosmic microwave background, the other by stars and supernovae. And the more precisely we compare them, the more their readings diverge. Now, a new 'clock' using only pulsating Cepheid stars has added a third, even more unexpected result.
Astronomers carefully accounted for the motion of galaxies relative to each other and obtained a Hubble constant of 71.1 (km/s)/Mpc. This is lower than the previous standard (73.2 from Adam Riess), but the discrepancy with the 'relic radiation' clock has grown to a level where chance is almost ruled out (less than 0.5%). Contrary to hopes, cleaning up the mechanism did not synchronize the clocks but only highlighted the fault.
In 1929, Edwin Hubble using the same Cepheids calculated 500 (km/s)/Mpc — it turned out he confused two types of these stars. We no longer confuse them, but the divergence hints that in the very mechanism of the cosmic clocks — the standard model — there may be some missing gears.
Edwin Hubble's first measurement in 1929 gave about 500 km/s per megaparsec — seven times larger than today's, because he didn't know about the two types of Cepheids.