Who measured the universe?

Henrietta Swan Leavitt found the relation between a variable star’s period and brightness. Her paper gives that relation, not a distance estimate. Ejnar Hertzsprung applied it to the Small Magellanic Cloud; Edwin Hubble used a calibrated version to estimate Andromeda’s distance. The source record is in separate papers: Ejnar Hertzsprung, ‘Über die räumliche Verteilung der Veränderlichen vom δ Cephei-Typus,’ Astronomische Nachrichten 196 (1913) comes before Edwin P. Hubble, ‘Cepheids in Spiral Nebulae,’ The Observatory 48 (1925).

Papacosta’s account in the American Astronomical Society’s STATUS newsletter places Leavitt at Harvard under director Edward Pickering. The women known as computers examined photographic plates, identified stars and catalogued them; Papacosta says they were paid much less than men. A plate left a record of a star field that could be examined after the night of observation. For a variable star, the repeating change matters as much as the measured brightness: the time between repetitions is its period. Leavitt used that kind of record to identify a pattern later astronomers could calibrate. That division of work is why the source trail matters: the relation and the distance calculations appear in different papers.

In her catalogue of Magellanic variables, Henrietta Leavitt studies stars in the Small Magellanic Cloud. Her record of the table reads: Table VI has 16 Small Magellanic Cloud variables. Its shortest listed period is 1.25336 days (No. 1505); its longest is 127 days (No. 821). On p. 107 Leavitt writes that “the brighter variables have the longer periods.”

Because the stars belong to the same cloud, differences in their apparent brightness can be compared without solving for each star’s distance. Leavitt notes that brighter variables have longer periods. A Cepheid’s period is the time its brightness takes to repeat; photographic magnitude is the scale used to record how bright it appears. The graph here uses the values Leavitt printed at maximum and minimum brightness as separate series. Its points let the reader see the pattern before the fitted lines appear. The relation is useful because a star’s cycle can indicate its intrinsic brightness; compare that with how bright it appears from Earth, and an observer can estimate distance. Leavitt establishes the relation; the distance calculations appear in Hertzsprung’s and Hubble’s papers.

The Harvard College Observatory circular carrying Leavitt’s period statement credits the work to her in its opening sentence: The circular states: “The following statement regarding the periods of 25 variable stars in the Small Magellanic Cloud has been prepared by Miss Leavitt.” The table supplies the observations behind the period–brightness relation. A relation is a rule, not a distance. To turn it into a distance, astronomers needed to assign an absolute brightness to stars with a given period. The graph makes Leavitt’s result inspectable: each star is plotted by its period and photographic brightness, with maximum and minimum observations kept separate.

A least-squares fit gives a slope of 0.46 for maximum-light measurements and 0.45 for minimum-light measurements; Leavitt’s circular prints about 0.48. These are the calculated values and the printed estimate, respectively. The graph fits log period against brightness, the direction Leavitt describes. Reversing the axes answers a different question and changes the slope, even when the same stars remain on the plot. The calculation uses the transcribed table rows; it is separate from the approximate value in the circular.

In his paper on Cepheids, Ejnar Hertzsprung applies Leavitt’s relation to variables in the Small Magellanic Cloud. The step from a pattern to a distance needs a zero point: the absolute brightness assigned to a star with a given period. Once that is known, the star’s observed brightness can be compared with its expected brightness at a standard distance. Hertzsprung’s paper preserves a problem worth keeping visible. Its printed values do not agree: On p. 204 Hertzsprung gives the mean absolute brightness as −7.3 for a 6.6-day period; the Small Magellanic Cloud variables of that period have mean photographic magnitude 14.5, and with colour index +1.5 he takes the visual magnitude as 13.0. He prints 5 log p = −7.3 − 13.0 = −20.3, p = 0.0001 arcsec, corresponding to about 3000 light-years.

The mixed-input diagnostic is 241,571 pc; it is not a distance estimate. It applies Shapley’s visual zero point and Leavitt’s photographic slope to Hubble’s photographic maximum without converting the visual curve or correcting for the star’s phase. Hubble converted Shapley’s curve to photographic magnitudes and corrected maximum light by half the average range before reporting about 285,000 parsecs from 12 Cepheids in Andromeda. The calculator therefore shows only a magnitude comparison factor against the Small Magellanic Cloud; its mean-magnitude reference and Hubble’s maximum-light input do not make a distance ratio. Riess et al.’s modern Hubble Space Telescope (HST) calibration uses a Wesenheit magnitude, a reddening-corrected combination of measurements through several filters; its absolute magnitude is -5.839 at a reference period of 10 days. Its filters and mean-light measurements do not match Hubble’s photographic maxima, so no distance is calculated from it.

The National Research Council’s report of the Shapley–Curtis debate preserves the speakers’ positions. Harlow Shapley argued for a Milky Way far larger than the model then accepted and placed the Sun away from its centre; he treated spiral nebulae as objects within the Milky Way. Heber Curtis defended the smaller, Sun-centred model and argued that spiral nebulae were separate galaxies. The report records the cases without a retrospective verdict. The disagreement mattered because the same spiral could be read as a nearby part of the Milky Way or a stellar system beyond it.

Read beside Hubble’s paper on Cepheids in Andromeda, Curtis’s separate-galaxy interpretation has support: Hubble’s distance estimate places Andromeda beyond the Milky Way. This comparison uses the debate report and Hubble’s paper; the report itself does not name a winner.

Carnegie Science’s plate archive records Hubble’s observation of a changing star in Andromeda: Carnegie Science’s plate archive reproduces Hubble’s H335H (“Hooker plate 335 by Hubble”) and says it was taken on the night of 5–6 October 1923. The plate image shows the first Cepheid’s “N” crossed out and “VAR!” written beside it. The archive dates the exposure to Night of 5–6 October 1923. The plate records an annotation; Hubble’s distance estimate appears in a separate published paper.

In *Cepheids in Spiral Nebulae*, Edwin Hubble tabulates periods and photographic magnitudes for Cepheids in Andromeda, then derives a distance of 285,000 parsecs. That is the estimate shown on the ruler. Leavitt’s period–brightness relation supplies an estimate of intrinsic brightness; Hubble compares it with how faint the stars appear from Earth. His calculation puts Andromeda beyond the Milky Way, supporting the conclusion that spiral nebulae can be separate galaxies. The paper’s table and distance are the evidence for that claim. The later history of cosmology should not be read back into the plate annotation.

In Hubble’s velocity–distance paper, radial velocity means motion along the line of sight. The paper fits nebular distances against those velocities and adopts a representative constant: For 24 individual nebulae K = +465 (50); for nine groups K = +513 (60), in km/s per 10^6 parsecs. He adopts K = +500 km/s per million parsecs as a rounded representative value. In its closing paragraph, Hubble calls the relation roughly linear and says it is too early to discuss the consequences in detail: He concludes that the results establish a roughly linear velocity-distance relation, then says it is premature to discuss its obvious consequences in detail. He calls the outstanding feature the possibility that the relation may represent the de Sitter effect. The paper does not use the phrase ‘expanding universe’ or state that conclusion. The paper does not describe the result as an expanding universe.

Georges Lemaître made a separate calculation from listed nebulae. The English translation omits numerical details present in the optical-character-recognition transcription of his original: Lemaître says he uses 42 nebulae listed by Hubble and Strömberg: mean distance 0.95 million parsecs and mean radial velocity 600 km/s, equivalent to 625 km/s per million parsecs. Without weighting the observations, he gives 870 km/s at 1.16 million parsecs, or 575 km/s per million parsecs. He adopts R′/R = 0.68 × 10⁻²⁷ cm⁻¹. The scanned pages could not be checked, so the figures remain unverified against the original images.

Walter Baade’s account of his Cepheid revision describes its effect on distances based on Type I Cepheids: Baade's published account says he presented the results at the IAU's Rome meeting in September 1952. The revised zero point leaves distances based on cluster-type variables unchanged; distances previously derived from Type I Cepheids must be multiplied by 2, because those Cepheids are 1.5 magnitudes brighter than previously thought. This corrects the luminosity calibration; it is not a remeasurement of Hubble’s plate. The ruler’s ghost mark shows the result of applying Baade’s reported correction to Hubble’s published Andromeda distance. That mark is calculated from the source papers, rather than printed as a new distance in Baade’s account.

Li and colleagues used Cepheids observed with the Hubble Space Telescope and calibrated against Cepheids in the Large Magellanic Cloud (LMC), using the same filters, to measure the distance to Andromeda: 761,000 ± 11,000 parsecs. This modern mean-magnitude Wesenheit measurement sits beside Hubble’s estimate and Baade’s correction; each mark identifies its source and calibration.