Ground Truth · the velocities were Slipher's
The Redshift Before the Law
Ask the internet who discovered the expanding universe and it answers, fast and wrong-ish: Edwin Hubble, 1929. But the velocities that made Hubble's famous plot were measured by Vesto Slipher more than a decade earlier — and Hubble's paper names no source for them. Slipher's very first galaxy, the Andromeda nebula in 1913, isn't even receding: it's a blueshift, coming toward us at −300 km/s. Recompute that shift yourself, from the same calcium lines his spectra rode on.
The whole "discovery" hangs on measuring a wavelength nudge of about four ångström — one part in a thousand. Drag the velocity below and watch the two calcium absorption lines (Ca II H & K) slide off their laboratory positions. Every number is computed on the spot from v = c · Δλ / λ, the same arithmetic the verifier runs offline — nothing here is pre-baked.
Slipher's plates of Andromeda gave −284, −296, −308 and −301 km/s; he published the round figure −300. Blueward is approach, redward is recession.
This runs the formula backwards: v = c · (λ_obs − λ_rest) / λ_rest. Typing here moves the slider — the two views are the same equation.
At −300 km/s the shift is tiny: −3.94 Å on a 3933.66 Å line, a redshift/blueshift of z ≈ −0.10%. That "smudge to the blue" is the entire signal. Slipher's genius was seeing it at all in 1912 with a hand-guided glass plate and a nights-long exposure — the first radial velocity ever measured for a galaxy.
The check — every number recomputed in front of you
Live, for the velocity you've set (−300 km/s): the shift of each calcium line, from Δλ = λ·v/c with c = 299 792.458 km/s and the air rest wavelengths.
The one genuinely independent check
Slipher published a velocity, not raw line positions — so the recompute above is a faithful reconstruction. The real test is whether his 1913 number survives a century:
| Slipher 1913 (M31) | −300 km/s (plate mean −297.25, rounded) |
| Modern SIMBAD/NED | −301 ± 1 km/s → agrees to within 1 km/s ✓ |
Even the famous number was wrong — the scale-of-error check
The slope everyone credits to Hubble was itself off by nearly an order of magnitude, because the distance ladder of the 1920s was badly miscalibrated:
| value | rate (km/s/Mpc) | vs modern H₀ ≈ 67–73 |
|---|---|---|
| Hubble 1929 | ≈ 500 (K = 465±50, 513±60) | ≈ 7.1× too high |
| Lemaître 1927 | ≈ 625 (575 unweighted) | ≈ 8–9× too high |
| modern | 67–73 | ≈ 1× (by definition) ✓ |
Run it yourself: node research/who-discovered-the-expanding-universe/verify-who-discovered-the-expanding-universe.mjs (all checks recomputed offline, no dependencies).
One measurement is not the discovery
Here is the honest part the popular story flattens. Andromeda is blueshifted — it is falling toward the Milky Way, pulled by the Local Group's own gravity. It shows nothing about cosmic expansion, and Slipher could not have known that when he measured it. The expansion signal is statistical: by 1917 he had 25 spiral-nebula velocities, and 21 of them were redshifts (only 4 approaching), averaging ~570 km/s — Slipher's own figure, "about thirty times the average velocity of stars." A lopsided catalogue of receding galaxies — that is the fingerprint, not any single object.
And Slipher read his shifts as ordinary Doppler motion, not as space itself stretching. The idea that the redshifts are the expansion of space belongs to Georges Lemaître, who derived the linear velocity–distance law from general relativity in 1927 — two years before Hubble — resting the empirical side on Slipher's and Hubble's own velocity lists. So the credit splits three ways: Slipher measured, Lemaître explained, Hubble supplied the distances and the plot.
- Sept 1912Slipher obtains the first spectrogram of the Andromeda nebula (M31) at Lowell Observatory.
- 1913Slipher publishes M31 at −300 km/s (approaching) — Lowell Obs. Bull. No. 58, the first galaxy radial velocity, "the greatest hitherto observed."
- 1914–15~15 nebulae measured (~11 redshifted); the American Astronomical Society gives Slipher a standing ovation.
- 1917Slipher's review: 25 velocities, 21 receding / 4 approaching, average ~570 km/s — the expansion fingerprint, unread.
- 1922Friedmann finds expanding solutions of general relativity — the mathematics, ahead of the data.
- 1927Lemaître derives the linear velocity–distance law from relativity and estimates ≈ 575–625 km/s/Mpc (625 with weighting) — two years before Hubble.
- 1929Hubble supplies distances and the observational law (K ≈ 500); 46 velocities, only 24 with individual distances — and cites no source for the velocities except Humason (4 recent ones).
- 1931Lemaître himself drops his rate derivation from the English translation of his 1927 paper (Livio 2011) — not censorship, his own choice, deferring to Hubble's stronger results.
- 2018The IAU votes 78% in favour of renaming the relation the Hubble–Lemaître law.
What's proven here, what's cited, and every free choice in the recompute
Exactly, arithmetically true (recomputed live and offline). The Doppler relation v = c·Δλ/λ; for M31's published −300 km/s that gives z = −1.0007×10⁻³, Ca K shifting −3.936 Å (to 3929.72 Å) and Ca H −3.971 Å (to 3964.50 Å). The reverse control inverts the same equation. Hubble's own paper states the 46/24 split and K = 465±50, 513±60 (he proposed 500).
A reconstruction, not a re-derivation. Slipher's 1913 paper reports reduced velocities, not per-line Δλ from his plate micrometer. So "recompute the shift from a spectral line" faithfully reconstructs what a −300 km/s Doppler shift does to the Ca H&K lines he measured "over the region from F to H" — it is not a re-reading of his glass plates. The genuinely independent check is his −300 km/s against the modern −301 ± 1 km/s (SIMBAD/NED).
Free choices, and how little they matter. Ca K vs Ca H differ by ~0.9% in shift (both ≈ 4 Å); air vs vacuum rest wavelengths differ ~0.03%; Slipher's rounded −300 vs the unrounded plate mean −297.25 differ ~0.9%; classical vs relativistic Doppler differ ~0.05% (≈0.15 km/s) at this z — smaller than his own plate-to-plate scatter. None changes the story.
Cited, not recomputed. Lemaître's 1927 derivation and its rate — ≈ 625 km/s/Mpc with weighting, ≈ 575 unweighted (van den Bergh 2011 foregrounds 575); Friedmann's 1922 solutions; the 1931 translation omission (Mario Livio, Nature 479, 171, 2011, who showed Lemaître himself removed the paragraph); the 1914 AAS ovation and ~15-nebula count; the IAU's 2018 78% vote. Modern H₀ is quoted as a range (67–73 km/s/Mpc) precisely because it is still contested.
What we lean on but didn't measure. The Sombrero (NGC 4594) ≈ +1000 km/s preset is well attested but less tightly pinned than M31, so the recompute is anchored on M31, whose primary velocity table we read directly. The ~570 km/s average for the 1917 set is Slipher's own stated figure ("the average velocity 570 km… about thirty times the average velocity of stars"), quoted rather than re-reduced line-by-line here. (The ~700 km/s sometimes attached to this set is a different quantity — Slipher's derived velocity of the Sun's/Galaxy's own drift relative to the spirals, "about 700 km" toward RA 22ʰ — not the nebular mean.)