Annu. Rev. Astron. Astrophys. 1999. 37: 445-486 Copyright © 1999 by Annual Reviews. All rights reserved

2.2. Baade's Program Proposed

Hubble's distances to M31, M33, and NGC 6822 depended both on the absolute magnitude calibration of the P-L relation of classical Cepheids and the accuracy of their apparent magnitudes. Both parts of this dependence were uncertain in 1948.

Baade discussed both of these problems in his paper at the July 1, 1948 scientific dedication of the 200-inch. Among other points he said:

"Our first concern will be the extension of the magnitude scales down to magnitude 22.5, the limit attainable with the 200-inch. ... With the photoelectric cell at the 200-inch, we...shall be able to reach magnitude 18.5, leaving to photographic photometry the faint end from 18 mag, to 22.5 mag. which can be comfortably bridged in a single step."

The most fascinating part of his lecture is:

"Unfortunately...there remains considerable uncertainty regarding the adopted zero point of the period-luminosity relation of the Cepheids. ...We intend to derive the zero point both of the Cepheids and the cluster-type variables in an entirely [new] manner. There is every indication that beginning with spectral type F the dwarf branches in the populations I and II coincide. To mention just one piece of the evidence: the dwarfs nearer than 10 parsecs with well determined trigonometric parallaxes are scattered along the same line in the Hertzsprung-Russell diagram whether they are slow moving stars or are of the high-velocity type. Hence by extending the H-R diagram in a globular cluster like Messier 3 to dwarfs of solar brightness we shall be able to connect directly the cluster-type variables with G dwarfs, the mean absolute magnitudes of which are ultimately based on well-determined trigonometric parallaxes. We thus obtain the mean absolute magnitude of the cluster-type variables. Now with the 200-inch we shall be able to reach in the Andromeda nebula the cluster-type variables (and the long-period variables associated with the population II). Since we know their absolute magnitudes, a comparison with the classical Cepheids in the outer spiral regions will furnish the absolute magnitude of the latter."

"...Now we do not expect that this program, simple and straightforward as it may seem, can be carried out within the next three to four years, if necessary by resorting to so-called hard work."

In this last sentence, Baade was pointing out the importance of rare excellent seeing, rare enough that the work will be slowed because of it. But what actually happened is that the precept that the population I and II main sequences coincide (which was assumed by Baade) is not correct because of chemical composition differences - a difference not dreamed of in 1948.

When variations in the chemical compositions of stars did become apparent, a new program was required to solve the many problems caused by such variations. Those solutions were to occupy us for the first 20 years, and yet today, certain of the calibrations are still in doubt (many would say controversial). They include (a) the precise position of the main sequence as a function of metallicity, (b) the putative effect of chemical composition variations on the absolute magnitudes of the classical Cepheids (Sandage, Bell, Tripicco 1999), and (c) even yet, the absolute magnitude calibration of the RR Lyrae variables (Sandage 1993a, 1993b;, Reid 1999).

The story I tell next is how Baade (1952) concluded even from his first observations of M31 with the 200-inch that the Cepheid P-L relation that was first calibrated by Hertzsprung (1913), copied by Shapley (1918), used by Hubble (1925, 1926a, 1929a), and apparently confirmed by Wilson (1939), was undoubtedly too faint by about 1.5 magnitudes.

2.2.2. Baade's Program Realized

The four parts of the program outlined by Baade (1948) were (a) redetermine the magnitude scales to m = 22.5, (b) recalibrate the Cepheid P-L relation by the new method that was, (c) to find the main sequence of globular clusters and calibrate therewith the absolute magnitudes of RR-Lyrae variables, and (d) discover RR-Lyrae variables in the disk of M31 to compare with the apparent magnitudes of the M31 classical Cepheids, thus calibrating them.

All parts of this program were eventually carried out using both the Palomar and Mount Wilson telescopes. Much was learned in the process about stellar evolution, the chemical evolution of the Galaxy, methods to age-date the stars, and the relation between kinematics and chemical composition among the different stellar populations.

First Revisions of the Magnitude Scales Made at Mount Wilson     In the 1930s Baade had been engaged in redetermining the faint magnitude scales in particular Selected Areas of the original Kapteyn program (cf. Blaauw & Elvius 1964). He had found significant magnitude scale corrections to the Mount Wilson Catalog of Photographic Magnitudes in Selected Areas (Seares et al 1930). However, although Baade's photographic photometry had shown the urgent need for corrections to the Mount Wilson Catalog fainter than m_pg = 16, more accurate results could be obtained using the new photomultiplier cells developed during the war. At the request of Baade and Hubble, Stebbins, Whitford, and Johnson (1950) began a photoelectric campaign to test the faint magnitude scales of the Mount Wilson Catalog and the North Polar Sequence (Seares 1922).

Their important 1950 paper had a profound effect in showing that the Mount Wilson scales needed corrections at the faint end beyond m = 16 that sometimes reached 1.5 mag at m_pg (Seares) = 19. The result was a sober realization that a vast work in precision photometry for cosmology and stellar population studies lay ahead. That work, once begun, was to occupy the next 30 years.

Later Revisions of the Magnitude Scales at Mount Wilson and Palomar     A program of photoelectric photometry to determine faint magnitude sequences was begun soon after the 200-inch was operational. W.A. Baum, an excellent experimental physicist, was appointed to the Carnegie staff to begin the work. Baum (1946) had worked in the rocket group led by Richard Tousey at the U.S. Naval Research Laboratory. With this experience and the photometric expertise he had gained in his Caltech PhD thesis in measuring the extinction of the atmosphere near 3000 A cutoff, Baum came to the Observatories as the resident photometrist.

He supervised the construction of a prime-focus focal plane, pulse-counting photometer (Baum 1953) and started a program of three-color photometry of particular Selected Areas that were of importance for the galaxy programs of Hubble and Baade. Eventually, photometric sequences were determined to B = 22, well beyond the limit of the Mount Wilson Catalog.

Although never published, Baum's magnitude sequences in SA 61, SA 68, and SA 57 were used internally for many years at the Observatories for various programs where photographic transfers were still used to calibrate particular fields.

However, after about 1970 the observing programs that required photometry were planned around the modern practice of establishing photoelectric sequences directly in the fields of interest. Nevertheless, it is still of interest to determine the general level of corrections needed to the Mount Wilson Catalog even as bright as m = 18. To this end, the Selected Area program begun by Baum was continued into the 1980s using both the Mount Wilson and Palomar telescopes. The results for the 11 Selected Areas of SA 28, 29, 45, 55, 57, 71, 82, 94, 106, 107, and 118 are now in the literature (Sandage 1999).

The two main reasons why Hubble's extragalactic distance scale needed the large revision that is now nearly completed were (a) all his magnitude scales fainter than m_pg = 18 were too bright by as much as 2 mag at his limit, and (b) his adopted Cepheid P-L calibration was too faint by 1.5 mag.

2.2.3. Baade's Factor of Two in the Distance Scale

Failure to Resolve the RR Lyraes in the Disk of M31     Hubble's (1929a) distance modulus of M31 was m - M = 22.0. Baade (1944a) revised Hubble's modulus to m - M = 22.4, based on his preliminary revision of the Mount Wilson Catalog magnitudes in Selected Area 68, but still using Hubble's adopted zero-point of the Classical Cepheids.

The assumed absolute magnitude of RR Lyrae variables was M_pg(RR) = 0.0 at that time, based on the most recent calibration by Wilson (1939). This apparently confirmed an earlier statistical parallax determination by Bok and Boyd (1933). Wilson's analysis combined the more recent and more accurate proper motions with radial velocities of 55 RR Lyraes measured by Joy at Mount Wilson. Therefore, the calibration by Wilson of both the classical Cepheids and the RR Lyrae absolute magnitudes seemed secure.

Consequently, if m - M = 22.4 for M31, and with M(RR) = 0.0 for the RR Lyraes, Baade should have resolved RR Lyrae variables in the disk of M31 near the 200-inch plate limit of m_pg = 22.5. Even more telling, the giant branch of the M31 globular clusters should resolve easily beginning at m_pg = 21. Their absolute magnitudes were believed to be M_pg = -1.5 (M_vis = - 2.8) according to the HR diagrams known at the time (Baade 1944a, Figure 1).

What Baade actually observed is described in the report of IAU Commission 28 for the Rome meeting (Hoyle 1952):

"Baade then went on to describe several results of great cosmological significance. He pointed out that, in the course of his work on the two stellar populations in M31, it had become more and more clear that either the zero- point of the classical cepheids or the zero-point of the cluster variables must be in error. Data obtained recently - Sandage's colour-magnitude diagram of M3 - supported the view that the error lay with the zero-point of the classical cepheids, not with the cluster variables."

Following a request by Shapley for Baade to expand on the evidence, Baade replied, again at the Commission 28 IAU session:

"According to the present zero-points we should expect to find the cluster variables of the Andromeda nebula at m_pg = 22.4 since the distance modulus of this system, derived from the classical cepheids, is m - M = 22.4. The very first exposures on M31, taken with the 200-inch, showed at once that something was wrong. Tests had shown that we reach with this instrument, using the f/3.67 correcting lens, stars of m_pg = 22.4 in an exposure of 30 min. Hence, we should just reach in such an exposure the cluster-type variables in M31, at least at their maximum phases. Actually we reach only the brightest stars of population II in M31 with such an exposure. Since, according to the latest colour-magnitude diagrams of globular clusters, the brightest stars of population II are photographically about 1.5 mag. brighter than the cluster-type variables, we must conclude that the latter are to be found in M31 at m_pg = 23.9, and not at m_pg = 22.4 as predicted on the basis of our present zero-points."

"We have also convincing proof that the brightest stars of population II in M31 are properly identified because when they emerge above the plate limit the globular clusters in M31 begin to be resolved into stars. It should be emphasized that these are rough first data indicating the order of corrections which the present constants require."

To make these points secure was to occupy the next 20 years at Palomar.

Although Baade's public announcement was not made until the Rome meeting of the IAU in 1952, Hubble (1951) had already discussed Baade's discovery in his Penrose Lecture before the American Philosophical Society in April 1951.

He wrote:

"But now we find that something is wrong somewhere. The luminosities assigned to the Cepheids led us to expect that the globular clusters in M31 would be readily resolved with the 200-inch, and that their brightest stars, as well as comparable stars in the main body of the nebula, could be studied individually with ease. It was found, however, that the stars in question were fainter than expected by a considerable fraction of a magnitude, and that they could be recorded and studied only with difficulty. The discrepancy is important because M31 has been explored on the basis of Cepheids while our system has been explored, in a sense, on the basis of globular clusters."

The Route Through Globular Clusters: Discovery of the Main Sequence     Baade organized the work in two parts. First he began a long campaign on the Cepheids in M31 itself, mounting a discovery program for variable stars in four fields at varying distances from the center. He invited Henrietta Swope, a Harvard expert in photographic photometry, to join him in the analysis of his 200-inch plates. A first summary of the Cepheids found in the inner three fields is given by Baade & Swope (1955).

By 1963 Swope had completed the analysis of the crucial outlying Field IV (Baade & Swope 1963) where the absorption and crowding is minimal. Baade and Swope used a local photoelectric sequence that had been measured by Arp with the 200-inch and published as part of the Baade-Swope paper. A well-defined P-L relation was found in B and V. The mean magnitudes of the Cepheids ranged from B = 20.5 to 22.8. These data, together with the new Cepheid P-L calibrations by Arp (1960a) and by Kraft (1961), which we discuss in a later section, gave distance moduli of 24.84 in B and 24.68 in V. These are close to the modern value of (m - M)_o = 24.44 (Madore & Freedman 1991) as corrected for E (B-V) = 0.10 mag reddening.

The second part of Baade's program was to tie the classical Cepheids to the RR Lyraes via observations in the disk of M31. As was discussed earlier, this required discovering the main sequences of globular clusters and fitting the resulting HR diagrams with the main sequence stars with known trigonometric parallaxes. The vertical fit of the fitted HR diagrams calibrated the luminosities of all the cluster stars, including the RR Lyrae variables.

The latter was the problem Baade posed to Arp and this writer, his two graduate students at the time. Baade proceeded to train us well in the then extant large-telescope techniques of observing on the Newtonian platforms of such telescopes as the Mount Wilson 60-inch at first, and later at the 100-inch Hooker reflector. Using plates of M92 calibrated by photographic transfers to Selected Areas 61 and 57, close to the two target globular clusters of M92 and M3, we just reached the main sequence in M92 near V = 19.

Near the end of the two-year observing program we were joined by Baum who supplied a photoelectric calibration of our photographic sequences, and therefore of all the cluster stars in M92.

At the same time, Baade had taken a series of plates of the cluster M3 in B and V to carry the work to the limit of the Palomar telescope. The M3 main sequence was reached easily, and the photometry could be extended for several magnitudes beyond the turn-off point. With the main sequences in two globular clusters now identified, we were confident enough to announce the result at the Cleveland meeting of the American Astronomical Society at a symposium on the HR diagram (Arp, Baum, & Sandage 1952; ABS). The details for both clusters were published a year later (ABS, 1953, Sandage 1953).

We made the fit of the M92 and M3 main sequences to the main sequence of the nearby trigonometric stars in the manner suggested by Baade in his 1948 lecture. Our result that M_V(RR) = 0.0 seemed to confirm Baade's conclusion, stated in Rome, that the error was not in Wilson's absolute magnitude of the RR Lyraes but rather that the distance to M31 must be larger than Hubble's (1929a) value. As a consequence the error must be in the adopted P-L relation of the Cepheids by the reasoning that Baade had set out at Rome. Baade's result is generally considered to be the first major discovery with the Palomar telescope.

But convincing as our main sequence fits appeared to be, the problem of calibrating the RR Lyrae variables was eventually found to be considerably more complicated. The positions of the population I and II main sequences in the HR diagram were soon shown not to be the same, contrary to Baade's assumption. Involved were (a) the central discovery of the chemical variations among the stars, (b) its meaning for stellar evolution and Galactic structure, (c) the relation of the discovery to the work by Burbidge et al (1957) on the formation of the chemical elements, and (d) the theoretical expectation and the observational proof that the luminosity of stars at a given temperature depends on the metallicity, i.e. that main sequence positions are functions of [Fe/H].

This was all heady stuff at the time. The daily excitement, fervor, and transcendental character of the work, together with its promise for a cosmological synthesis of origins and astronomical evolution in the broadest sense, is impossible to describe. All who lived through the development from 1950 well into the 1980s will recognize these years as being as close to ineffable perfection as a scientific life can get.