The birth of the parsec
In short, “parsec” is a portmanteau-name (PARallax of one SECond of arc) proposed by Herbert Hall Turner in 1913 to characterize one unit of stellar distances. Actually, the history is slightly longer, as three different issues were interlaced at that time:
The unit that is used nowadays is thus the parsec, a distance at which the mean radius of the earth's orbit subtends an angle of one second of arc. The absolute magnitude M is defined as a function of the apparent magnitude m and distance d through M = m - 5 log(d/10 pc). But where these complicated definitions come from?
Sirius, the standard
Rather logically, some unit of stellar distance was used with the first studies about galactic astronomy done by the forerunner William Herschel at the end of the 18th century. When he began “ gauging ” our Galaxy to understand its shape, knowing precisely the distance of the stars was still out of reach (the first annual parallaxes were measured more than half a century later only). Herschel thus decided to use the distance of Sirius, as a distance unit being representative of the average distance of the first magnitude stars. This was under the hypothesis (though not correct, as the one assuming that all stars had the same absolute magnitude) “ that those of the second magnitude are at double, and those of the third at treble the distance, and so forth ” (Herschel, 1785). Incidentally, one should note that the distance of Sirius was not completely unknown at that time as Huygens in 1698, then more precisely Newton in 1685, had obtained an estimation (not that far from the actual value) through a photometric method; for the Herschel studies, Sirius only plays the role of a standard of distance.
The evolution of the minds
However, even before stellar distances could be unambiguously determined (around 1838 by Bessel, Struve and Henderson), it was known that the parallaxes were smaller than the arcsecond, thus showing how immense the distances from the sun to the stars were and how unhandy the usual distance units are. The speed of light is also known and, as written by Bode (1768), much more than the distance travelled at the speed of a cannonball, “ nature has a much larger measure to determine the prodigious distance of the fixed stars, without having to multiply numbers: this is the progression of light ”. He was not the first to use the light speed as distance unit: in 1694, Roberts writes “ that Light takes up more time in Travelling from the Stars to us, than we in making a West-India Voyage (which is ordinarily performed in six Weeks) ”.
Expressing stellar distances in light-year units was thus not unusual and this, even before the end of the 18th century. Later, when he lectured about the first determinations of the few (8) annual parallaxes known at that time (before 1846), Arago mentions their parallax, their distance from earth in astronomical units and league, but he also finds useful to write another table in years of light. Flammarion (1865) also gives an idea of the distances of stars using the time delay for light rays (“ light does not take less than 22 years to come from Sirius ”). As for the name “light-year ” itself, the moment where its first use occurs is not obvious. Although astronomy books seem to mention it at the end of the 1880s, it looks present in the usual vocabulary at least in 1868 (e.g. Fay, 1868).
A unit not really united
While light-year was being used, however, and much later (at the beginning of the 20th century), several scientific papers use various distance units, with various names. Kobold (1906) coins a “ Sternweite ” (stellar distance), the distance corresponding to a one arcsecond parallax. In 1909, Seeliger defines one “ Siriusweite ” (Sirius distance, a link with Herschel), as the one corresponding to a 0.2" parallax (i.e. 1.03 106 astronomical units), an unit he already used in his famous study about stellar statistics in 1898. Charlier (1913) mentions that he coined the word “ Siriometer ” since 1911, defining it as exactly 106 astronomical units. In 1912, Turner uses in the same article the light-year for the distance of A type stars (p. 482) and uses also the distance corresponding to a 1" parallax two pages later, without unit name. In short, there is much confusion.
This problem occurs at a moment in time where the proper motion data, and in a smaller extent of parallax data for which they are a substitute, allow the study of the “ structure of the universe ” as written by Kapteyn. His researches, together with Seeliger, and they are soon followed by others, intend to determine the spatial structure from stars of various spectral types, proper motions, absolute magnitudes and distances.
As indicated, the use of a distance unit, whatever its name, has an implication on the definition of the absolute magnitude. In order to be able to compare stars of various types, there is indeed a need to define the intrinsic luminosity one way or another. As the luminosity decrease with the square of distance and the apparent magnitude is defined as 2.5 times the logarithmic scale of the luminosity, the difference between apparent and absolute magnitude is 5 times the logarithm of distance plus a constant,... which depends on the adopted distance unit. For example, when Hertzsprung (1913) compared the intrinsic luminosities of stars in his cepheid calibration, he assumed 1 parsec as reference, as did Eddington (1914). On the contrary, Plummer (1912) uses an “ intrinsic magnitude ” with 100 pc as reference. In between, and for several years however, Kapteyn (1902, 1910) has clearly introduced the definition still in use today… but this was done consistently with a distance unit equal to 10 pc. Missing some coordination, results are quite difficult to compare.
The origin of the word “ parsec ” dates from this epoch. In his study where he used distances with units corresponding to one arcsecond parallax, Dyson (1913) indicates in a footnote that he suggests “ astron ” as unit name, while Turner had coined “ parsec ”, probably after reading the first version of this text. This triggers debate. The discussion occurs during the presentation of the paper at the Royal Astronomical Society (Dyson, 1913b),: Turner (p. 167) comes back on the designation of the distance unit. He estimates that “ astron ” sounds like “ astronomical unit ”, and (after having proposed “ parsec ”) he now also suggests “ macron ”. Dyson (p. 168) answers that “ Siriometer suggests a machine for measuring and then thought that the meaning of micron may lead to confusion.
There is no consensus yet. Paddock (1913) summarizes the current status, adding that “ In discussing distances of nebulae, Professor Very, proposes as a unit the distance of the Andromeda nebula and the name '' andromede '' ”. Among all these suggestions “ the designation astron, or perhaps astrometer, would seem very appropriate ” to him.
In response, Curtis (1913) devotes an article in the same volume (PASP) about this issue. He contends that an unit based on fundamental constants (such as the speed of light) would be preferable. Also, he notes that none of the proposed units is perfectly known: the astronomical unit not better (at that time) than the thousandth, the Sirius parallax is between 0.34 and 0.40; as for Andromeda, distance is not known to better than 50% (actually,it was 160000%!). Curtis also suggests that an unit should be useful to specialists as well as to the layman, and that he “sees no advantage in referring to the distance of a star as 14.3 parsecs, instead of saying that the star has a parallax of 0.07 seconds of arc ”. At the same epoch as the well-known debate about the size of our Galaxy, Curtis adds “whether the Milky Way is three thousand, or several million light-years across, the light-year is still a usable unit ”.
Eddington (1914) does not share this opinion and regrets that “ light-year which, notwithstanding its inconvenience and irrelevance, has sometimes crept from popular use into technical investigations ”. Finding a serious motive for the rejection of light-year is difficult, the only argument is given by Dyson (1913b) who “ feels that the use of light-years has been introduced for lecture purposes, not for operations when your lecturer is working with the star ”.
The I.A.U., a normalisation organisation
The parsec looks quickly adopted by Eddington (1914) in his book about stellar motions. In 1919, the (new, and now vanished) Commission “ Notations ” of the International Astronomical Union (IAU) recommends the use of light-year “ especially in popular articles ”, du parsec, “ or preferably a unit 10 times as great, to be given a separate name ” (Campbell, 1920). However, at the same epoch, Kapteyn (1920) now acknowledges the existence of parsec “ which is very convenient (though very ugly) ” and now intends to define the absolute magnitude using one (not 10) parsec units! In 1922, the IAU Commission 3 decides the use of m and M for the apparent and absolute magnitudes respectively, and the use of the parsec, without the noted restriction. In 1925, the AU confirms these units. Which does not mean an overwhelming agreement: Malmquist (1925) rightly notes that the distance unit is the parsec while the absolute magnitude is defined with respect to a multiple of the parsec… And, in the same volume of the Observatory, Doig (p. 138) express the dimensions of M31 and M33 or the distance to open clusters in light-years, while Hubble in his paper (p. 142) signing the “ end of Shapley's universe ”, begins to use parsecs, though noting its light-year equivalent.
At next IAU General Assembly, Charlier (1928) had not given up. Taking account of the reluctance for the word “ siriometer ” he had proposed before (he also mentions that some French astronomers had suggested “ Herschel ” as unit name), he suggests as stellar distance unit: 106 astronomical units, as time unit: 106 tropical year, and the absolute magnitude as the magnitude at this distance unit. As president of IAU Commission 33 of stellar statistics, he expects the vote of a resolution of his Commission towards the Commission of Notations (p. 260). He looks being supported by Malmquist only (who had been his student...). In front of the bright Lund school : Russel, Lundmark note the disadvantage of a modification of units now entered in wide use in the literature. Malmquist signals the contradiction between the definition of the parsec and the one of the absolute magnitude, but Eddington replies that physics uses the liter defined as a cubic decimeter... The Charlier proposal was thus rejected.
However, most of star distances are known indirectly only (not measured angularly from annual parallaxes). For instance, parallaxes of nearby stars allow calibration for several spectral types, and the subsequent use of these calibrations for distant stars give what is called “ spectroscopic parallaxes ”, a strange name indeed. Beyond, the gigaparsec is used while technology is far from allowing the measurement of annual parallaxes at the nanoarcsecond level! This is however the parsec, a unit which originates from the measurement method (such as the foot or inch), this “ portmanteau-name ” (Eddington) which remained in use. Perhaps also because it was a “ smart name ” (Malmquist). As for the absolute magnitude, the fact that it was the oldest definition, and used by most astronomers probably explains its use.
In summary, the “ parsec ” has first been used as unit by Kobold, its name coined by Turner, while Kapteyn is probably the first to have defined the absolute magnitude as it is used today.