<<

Hubble’s diagram and cosmic expansion

Robert P. Kirshner* Harvard–Smithsonian Center for , 60 Garden Street, Cambridge, MA 02138

Contributed by Robert P. Kirshner, October 21, 2003

Edwin Hubble’s classic article on the expanding appeared in PNAS in 1929 [Hubble, E. P. (1929) Proc. Natl. Acad. Sci. USA 15, 168–173]. The chief result, that a ’s is proportional to its , is so well known and so deeply embedded into the language of through the Hubble diagram, the Hubble constant, Hubble’s Law, and the Hubble time, that the article itself is rarely referenced. Even though Hubble’s have a large systematic error, Hubble’s come chiefly from Vesto Melvin Slipher, and the interpretation in terms of the de Sitter effect is out of the mainstream of modern , this article opened the way to investigation of the expanding, evolving, and accelerating universe that engages today’s burgeoning field of cosmology.

he publication of Edwin Hub- ble’s 1929 article ‘‘A relation between distance and radial T among extra-galactic nebulae’’ marked a turning point in un- derstanding the universe. In this brief report, Hubble laid out the evidence for one of the great discoveries in 20th cen- tury : the expanding universe. Hubble showed that recede from us in all directions and more dis- tant ones recede more rapidly in pro- portion to their distance. His graph of velocity against distance (Fig. 1) is the original Hubble diagram; the equation that describes the linear fit, velocity ϭ ϫ Ho distance, is Hubble’s Law; the slope of that line is the Hubble con- ͞ stant, Ho; and 1 Ho is the Hubble time. Although there were hints of cosmic Fig. 1. Velocity–distance relation among extra-galactic nebulae. Radial velocities, corrected for solar (but labeled in the wrong units), are plotted against distances estimated from involved and expansion in earlier work, this is the mean of nebulae in a cluster. The black discs and full line represent the solution for solar publication that convinced the scientific motion by using the nebulae individually; the circles and broken line represent the solution combining the community that we live in an expanding nebulae into groups; the cross represents the mean velocity corresponding to the mean distance of 22 universe. Because the result is so impor- nebulae whose distances could not be estimated individually. [Reproduced with permission from ref. 1 tant and needs such constant reference, (Copyright 1929, The Huntington Library, Art Collections and Botanical Gardens).] have created eponymous Hubble entities to use Hubble’s aston- ishing discovery without a reference to ridge, luminous reveals the pres- of acceleration set in are the route to the original publication in PNAS (1).† ence of unseen objects. illuminating this mystery. Extensions of Hubble’s work with to- Hubble applied the fundamental dis- Today, Ͼ70 years later, exquisite ob- day’s technology have developed vast coveries of Henrietta Leavitt concern- servations of the cosmic microwave new arenas for exploration: extensive ing bright Cepheid variable stars. background (2), measurement of mapping using Hubble’s Law shows the Leavitt showed that Cepheids can be elements synthesized in the first few arrangement of matter in the universe, sorted in by observing their minutes of the universe (3), and modern and, by looking further back in time vibration periods: the slow ones are versions of Hubble’s Law form a firm than Hubble could, we now see beyond the intrinsically bright ones. By mea- triangular foundation for modern cos- the nearby linear expansion of Hubble’s suring the period of pulsation, an ob- mology. We now have confidence that a Law to trace how cosmic expansion has server can determine the ’s intrin- geometrically flat universe has been ex- changed over the vast span of time since sic brightness. Then, measuring the panding for the past 14 billion yr, grow- the . The big surprise is that apparent brightness supplies enough ing in contrast through the action of recent observations show cosmic expan- information to infer the distance. from a hot and smooth Big Bang sion has been speeding up over the last to the lumpy and varied universe of gal- 5 billion yr. This acceleration suggests axies, stars, , and people we see that the other 70% of the universe is This Perspective is published as part of a series highlighting around us. Observations have forced us landmark papers published in PNAS. Read more about composed of a ‘‘dark ’’ whose this classic PNAS article online at www.pnas.org͞misc͞ to accept a dark and exotic universe properties we only dimly grasp but that classics.shtml. Ϸ that is 30% with only 4% must have a negative pressure to make *E-mail: [email protected]. of the universe made of familiar protons cosmic expansion speed up over time †There are just 73 citations of Hubble’s original paper in and neutrons. Of that small fraction of (4–9). Future extension of the Hubble NASA’s Astrophysics Data System. There are 1,001 citations familiar material, most is not visible. diagram to even larger distances and of ref. 7. Like a dusting of snow on a mountain more precise distances where the effects © 2003 by The National Academy of Sciences of the USA

8–13 ͉ PNAS ͉ January 6, 2004 ͉ vol. 101 ͉ no. 1 www.pnas.org͞cgi͞doi͞10.1073͞pnas.2536799100 Downloaded by guest on September 28, 2021 PERSPECTIVE

Hubble used the 100-inch Hooker of a and passing it through a which particles would scatter with an Telescope at Mount Wilson to search slit and a prism to create a dispersed acceleration increasing with distance for these ‘‘standard candles’’ and found rainbow, subtly marked by dark lines. and signals sent from one observer to Cepheids in the fuzzy Andromeda These absorption lines are produced by another would show a redshift. More Nebula, M31. From the faint appear- in the atmospheres of stars. At- physical solutions to Einstein’s equa- ance of those Cepheids, Hubble de- oms absorb light at specific , tions, constructed for an expanding duced that M31 and the other ‘‘extra- matching the energy jumps for electron universe, were worked out by Fried- galactic nebulae’’ are not part of our dictated by quantum mechanics. mann in 1922. But those were not the own galaxy, but ‘‘island uni- Radial velocities show up as shifts in the models Hubble was thinking of when verses’’ equivalent to the Milky Way: wavelengths of the lines from the galaxy he plotted his data. Hubble was look- vast systems of billions of stars sepa- compared with the spectra of the same ing for the de Sitter effect. rated from one another by millions of atoms at rest in the observatory: - Putting distances and velocities to- light years. This finding was in 1924, shifts for objects approaching us and gether on the graph shown as Fig. 1 in and if he had done nothing more than for objects receding. The frac- Hubble’s classic article, anybody can see to show that the Milky Way is not the tional shift of the , ⌬␭͞␭,is1 that the velocity is more or less propor- universe, Hubble would have been an ϩ z, where z is the redshift. This result tional to the distance. What transforms important figure in the history of as- can be expressed as a velocity, cz, where this bland diagram into a profound dis- tronomy. But 5 yr later in his PNAS c is the , 300,000 km͞s. covery is an understanding that the pat- article, Hubble was able to show some- The program of measuring galaxy tern Hubble found is exactly what you thing even more astonishing by plotting spectra had been initiated a decade ear- would expect for any observer in a uni- the velocities of galaxies against their lier by Vesto Melvin Slipher at the Low- verse expanding in all directions. Hub- distances. ell Observatory in Arizona. By 1923, ble’s diagram does not imply that we are Reading Hubble’s article is a healthy after heroic efforts with small at the center of the universe, but it does reminder of how much clearer things and slow spectrographs, Slipher had show that the universe is dynamic, defi- become with 70 yr of hindsight. For ex- compiled a list of velocities for 41 galax- nitely not static, as Einstein had as- ample, although the Cepheids lie at the ies, 36 of which were receding from us, sumed in 1917. foundation of Hubble’s distance scale, and the largest of which was moving In the text of his article, Hubble says, the distances to most of the objects in away at 1,800 km͞s. This intriguing list ‘‘the outstanding feature is the possibil- his 1929 article were not determined by was published in Arthur Stanley Edding- ity that the velocity-distance relation Cepheids themselves, but by the bright- ton’s textbook on , The may represent the de Sitter effect.’’ It is est stars in galaxies or by the luminosity Mathematical . probably not a coincidence that Hubble of the galaxies themselves. In recent Hubble cites no source for the radial was looking for this effect: he was in years, by using the superb resolution of velocities in table 1 of ref. 1, except for in 1928 for a conference on gal- the Hubble , named in the four new ones from his Mount Wil- axies, and he had the opportunity to ’s honor, it is finally possi- son colleague, Milton Humason, but talk with de Sitter. Hubble notes that ble to measure individual Cepheids in every one of the galaxies is one of one aspect of the de Sitter world model galaxies in the cluster that are the Slipher’s, and the list of velocities is al- is an apparent ‘‘acceleration,’’ and he most distant entries in Hubble’s original most identical to that in Eddington’s makes the plausible assumption that the table of galaxy redshifts and distances book. Hubble’s original contribution in ‘‘linear relation found in the (10, 11). The quantitative agreement of 1929 was to grasp the connection of dis- discussion is a first approximation repre- modern measurements with Hubble’s tance with velocity, and his subsequent senting a restricted range in distance.’’ original distance scale is not good! Mod- effort was to pursue the consequences This particular aspect of Hubble’s article ern distances to the same galaxies, reck- of this amazing fact. Hubble and Huma- has seemed quaint and puzzling. oned to be accurate to 10%, are seven son poured prodigious effort into mea- It has seemed quaint because, from times larger than the distances Hubble suring redshifts at the 100-inch, rapidly 1931 to about 1995, almost nobody was plots horizontally in Fig. 1. Hubble’s expanding the reach of the Hubble dia- talking about cosmic acceleration. As a essential contribution was a consistent gram beyond the 1,000 km͞s velocities result of Hubble’s own work, even Ein- set of distances to galaxies that allowed shown in Fig. 1. Although Slipher had stein and de Sitter stopped talking him to glimpse the underlying relation begun the field of galaxy spectra a de- about their old models with a cosmolog- between distance and velocity. Although cade earlier and measured the velocities ical constant, and the observational fo- his distances had serious errors due to that Hubble used in his 1929 article, cus shifted to finding the numerical confusing two types of Cepheids, and Hubble soon became the towering figure value of the Hubble constant and to blurring bright gas clouds with bright in exploring the realm of the nebulae. measuring the gravitational effect of stars, in 1929, Hubble was able to sort The connection between general rel- cosmic deceleration in an expanding nearby galaxies from distant ones well ativity and cosmic velocities was lurk- Friedmann model. And it seemed puz- enough not to miss the connection be- ing in the background of Hubble’s zling because, in modern parlance, an tween distance and velocity. work. In 1917, Einstein had shown how Einstein–de Sitter model is an expand- The other axis of the Hubble diagram to construct a universe that was static ing Friedmann model with a flat geome- (subtly mislabeled in the original) shows by introducing a ‘‘cosmological con- try, and the original de Sitter effect is a not only that we live in a spacious uni- stant’’ into his equations. This matched historical curiosity. But today, in light of verse populated by billions of galaxies well with the idea, current before Hub- recent work that suggests that we do like the Milky Way, but also that the ble’s 1924 measurement of the dis- live in an accelerating universe of the galaxies are embedded in an expanding tances to the nebulae, of a small and Einstein–de Sitter type, with Euclidean fabric of space and time. The Hubble static ‘‘universe’’ that was confined to space, Hubble’s allusion to acceleration diagram plots velocity against distance. the stars of the Milky Way galaxy. In seems oddly, perhaps falsely, prescient. Astronomers measure the velocity of a Leiden, had shown The slope of the line in the Hubble galaxy from its spectrum by taking the that there was another, formally static, diagram is called the Hubble constant light from a galaxy’s image at the focus solution to Einstein’s equations in (Ho), directly related to the age of the

Kirshner PNAS ͉ January 6, 2004 ͉ vol. 101 ͉ no. 1 ͉ 9 Downloaded by guest on September 28, 2021 universe: Hubble’s Law says that veloc- ϭ ϫ ity Ho distance and, because time ϭ distance͞velocity, there is a nat- ural Hubble time, to, associated with the ϭ ͞ Hubble expansion, to distance veloc- ϭ ͞ ϫ ϭ ͞ ity distance (Ho distance) 1 Ho. Nearby objects recede slowly, and more distant ones recede rapidly, but both would take the same time to get where they are in a universe that ex- pands at a constant rate, and that time ͞ is given by 1 Ho. So the Hubble con- stant sets the time scale from the Big Bang to today.‡ Although Hubble’s 1929 distances were too small by a factor of 7, his con- clusion about the of cosmic ex- pansion was still valid because all his distances were too small by about the Fig. 2. Published values of the Hubble constant vs. time. Revisions in Hubble’s original distance scale same factor. The form of the relation, account for significant changes in the Hubble constant from 1920 to the present as compiled by John velocity proportional to distance, is not Huchra of the Harvard–Smithsonian Center for Astrophysics. At each epoch, the estimated error in the changed by this scale error, although the Hubble constant is small compared with the subsequent changes in its value. This result is a symptom of numerical values for the distances, and underestimated systematic errors. for the Hubble constant (which Hubble modestly called K) is far from the mod- ern value. In this classic article, Hubble stant shows how the prevailing value has oldest stars in our galaxy have ages, quotes values of K of 530 and 500 been dropping over the decades. The based on computations of stellar evolu- km͞s͞megaparsec. Staring at his original quoted error bars are chronically much tion through nuclear burning, of Hubble diagram, you can see that there smaller than the drift in the mean value Ϸ12.5 Ϯ 1.5 billion yr, just enough is a handful of nearby galaxies with over time. The systematic errors are al- younger than the Hubble time to fit , and a large scatter of veloci- ways underestimated. This plot lends comfortably into a scheme where galax- ties at any given distance. Hubble weight to the aphorism that astrophysi- ies form promptly after the Big Bang shrewdly used plausible methods to av- cists are always wrong, but never in (17). Even with the added wrinkle of erage the data for galaxies that are at doubt. cosmic deceleration and cosmic acceler- the same distance to make his result Modern work, closely tied to the Cep- ation, the best value from the Hubble stand out more clearly from the noise. heids in galaxies observed diagram for the elapsed time since the He was fortunate to have data that be- with the gives Big Bang is Ϸ13.6 Ϯ 1.5 billion yr (18). ϭ Ϯ Ϯ ͞ ͞ haved so well. Ho 72 2 7km s megaparsec (9). The expansion is no illusion; it is cosmic Over time, improved understanding of The errors quoted are one sigma, with history. the stars being used, the role of absorp- the first being the statistical error, and As in Hubble’s original article, where tion by dust, and the local calibration of the second, larger error being the sys- he used the very brightest stars and the distance scale led to large revisions tematic uncertainty due to factors like the light from entire galaxies, the mod- in the cosmic distance scale, the Hubble the chemical composition of the Cep- ern path to deeper distance measure- constant, and in the inferred Hubble heids in different galaxies, the distance ments is through a brighter standard Ϸ time. In Hubble’s time, to was 2 billion to the Large Magellanic Cloud to which candle than the Cepheids. Before yr, which was already in conflict with the distance comparison is made, and Hubble, astronomers had, from time to the larger age of the Earth inferred the calibration of the camera on the time, noted new stars that flared up in from radioactive decay. The Earth Hubble Space Telescope. As in the past, extragalactic nebulae like M31 and its should not be older than the universe in we believe these error bars are correct cousins. In our own galaxy, these new which it formed. This conflict with the (although for a contrasting view, see ref. stars are called ‘‘novae.’’ Once Hubble age of the Earth and a similar problem 10). But now, the convergence from had established that the distances to with the ages of the stars was a chronic completely independent methods such these nebulae were millions of light embarrassment during the decades when as time delays in gravitational lenses, years, the true nature of these novae the Hubble constant was poorly known. of microwave background became clear. Because they were at The disagreement made it difficult to by hot gas in galaxy clusters, distances a thousand times larger than accept the reality of cosmic expansion and the of atmo- novae in the Milky Way, they must be acting over , and Hubble spheres is beginning to be significant a million times more energetic. Ex- was always quite circumspect on the in- (12–16). With independent paths, sys- ploding stars in other galaxies were terpretation of his discovery. But, as tematic errors can be exposed. We are, dubbed ‘‘supernovae’’ by , shown in Fig. 2 of this Perspective, my at last, coming to the end of the search Hubble’s contemporary down Lake Av- colleague John Huchra’s compilation of for the Hubble constant. enue in Pasadena at the California In- the numerical value of the Hubble con- The remarkable result of this long stitute of Technology. The light output path of revision is that the Hubble time of one particular type of supernova is is now taken seriously. The age of the Ϸ4 billion times that of the . These ‡The conventional units of the Hubble constant are a bit universe implied by the modern Hubble ‘‘type Ia’’ supernovae can be seen half obscure: 1 megaparsec (Mpc) ϭ 106 ϭ 3.26 ϫ 106 Ϸ light years ϭ 3.086 ϫ 1016 m. A Hubble constant of 70 constant with constant expansion is 14 way across the visible universe, and, km͞s͞Mpc corresponds to 2.27 ϫ 10Ϫ18 sϪ1. Then, the billion yr. This result is in good accord even better, they have a fairly narrow Hubble time is 1͞2.27 ϫ 1018 s or 13.9 ϫ 109 yr. with the theoretical ages of stars. The distribution in intrinsic brightness. As a

10 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.2536799100 Kirshner Downloaded by guest on September 28, 2021 letter to Willem de Sitter (25), Hubble wrote, ‘‘Mr. Humason and I are both deeply sensible of your gracious appreci- ation of the papers on velocities and distances of nebulae. We use the term ‘apparent’ velocities to emphasize the empirical features of the correlation. The interpretation, we feel, should be left to you and the very few others who are competent to discuss the matter with authority.’’ Part of the difficulty with the inter- pretation came from alternative views, notably by the local iconoclast, Fritz Zwicky, who promptly sent a note to PNAS in August 1929 that advocated thinking of the redshift as the result of an interaction between photons and in- tervening matter rather than cosmic ex- Fig. 3. The Hubble diagram for type Ia supernovae. From the compilation of well observed type Ia pansion (26). The reality of cosmic supernovae by Jha (29). The scatter about the line corresponds to statistical distance errors of Ͻ10% per expansion and the end of ‘‘’’ object. The small region in the lower left marks the span of Hubble’s original Hubble diagram from has only recently been verified in a 1929. convincing way. While the nature of the redshift was a bubbling discussion in Pasadena, Olin result, they make good distance indica- clustering leads to estimates for the Wilson of the Mount Wilson Observa- tors. Refined methods for analyzing amount of clumpy dark matter associ- tory staff suggested that measuring the the observations of type Ia supernovae ated with galaxies. The best match time it took a supernova to rise and fall give the distance to a single to comes if the clumpy matter (dark and in brightness would show whether the better than 10% (19, 20). The best luminous, baryons or not) adds up to expansion was real. Real expansion Ϸ modern Hubble diagram, based on well 30% of the universe. would stretch the characteristic time, observed type Ia supernovae out to a The interpretation of the redshift as a about a month, by an amount deter- Ϸ modest distance of 2 billion light velocity, or more precisely, as a stretch- mined by the redshift (27). years, is shown in Fig. 3, where the ing of wavelengths due to cosmic This was sought in 1974, axes are chosen to match those of expansion, which we assume today’s col- but the sample was too small, too Hubble’s original linear diagram (to lege sophomores will grasp, was not so nearby, and too inhomogeneous to see mask our uncertainties, astronomers obvious to Hubble. Hubble was very anything real (28). It was only with large generally use a log-log form of this circumspect on this topic and, more gen- carefully measured and distant samples plot as in Fig. 4). Far beyond Hubble’s erally, on the question of whether cos- of SN Ia (29, 30) and more thorough original sample, Hubble’s Law holds mic expansion revealed a genuine cos- characterization of the way supernova true. mic history. He referred to the redshift light curves and supernova luminosities In table 2 of his original article (1) as giving an ‘‘apparent velocity.’’ In a are intertwined (31, 32) that this topic (reproduced as Table 1, which is pub- lished as supporting information on the PNAS web site), Hubble inverted the velocity–distance relation to estimate the distances to galaxies of known red- shift. For galaxies like NGC 7619 for which he had only Humason’s recently measured redshift, Hubble used the velocity–distance relation to infer the distance. This approach to estimating distances from the redshift alone has become a major industry with galaxy redshift surveys. Today’s telescopes are 1,000 times faster at measuring red- shifts than in Hubble’s time, leading to large samples of galaxies that trace the texture of the galaxy distribution (21– 24). As shown in Fig. 5, the 3D distri- bution of galaxies constructed from Hubble’s Law is surprisingly foamy, with great voids and walls that form as dark matter clusters in an expanding Fig. 4. Hubble diagram for type Ia supernovae to z Ϸ 1. Plot in astronomers’ conventional coordinates universe, shaping pits into which the of distance modulus (a logarithmic measure of the distance) vs. log redshift. The history of cosmic ordinary matter drains, to form the expansion can be inferred from the shape of this diagram when it is extended to high redshift and luminous matter we see as stars in gal- correspondingly large distances. Diagram courtesy of Brian P. Schmidt, Australian National University, axies. Quantitative analysis of galaxy based on data compiled in ref. 18.

Kirshner PNAS ͉ January 6, 2004 ͉ vol. 101 ͉ no. 1 ͉ 11 Downloaded by guest on September 28, 2021 Supernova explosions behave better. As discrete physical events with a well defined energy, supernovae of type Ia work well as standard candles over a very large range in redshift. These ex- plosions allow us to look back to the time when the universe was young and to see the effects of changes in the rate of expansion reflected in the Hubble diagram. During the first stages of this work, the observers ex- pected to see the deceleration that would cause (39). The first re- ports from the Supernova Cosmology Project using supernovae confirmed Fig. 5. Large scale structure inferred from galaxy redshifts. Each dot in this plot marks a galaxy that view (40). However, better data whose distance is estimated from its redshift by using Hubble’s Law. From the 2DF Galaxy Redshift sets for the Hubble diagram of distant Survey (24). supernovae from the High-Z Super- nova Team and from the Supernova Cosmology Project (7, 8) showed the could be explored with confidence. Best of practical cosmology shifted to mea- surprising result that the expansion of of all was to have supernovae at high suring two parameters: the Hubble con- the universe has been speeding up dur- redshift where the effect would not be stant and the deceleration that gravity ing the 5-billion-yr interval while the subtle. Bruno Leibundgut et al. (33) produces over time. The goal was to light from a distant supernova has showed that the light curve for one ob- been in flight to our telescopes. ϭ construct a Hubble diagram in which ject, SN 1995K, at a redshift of z the most distant objects were sufficiently The most recent summary of the su- 0.479 matched the light curve of a far to show a clear deviation from the pernova data by Tonry et al. (18) nearby SN Ia, but only when stretched linear law seen by Hubble in 1929. In shows that this result is robust and fits by time dilation by a factor of 1 ϩ z. well with recent results on the micro- 1989, as in 1929, the problem was not Similarly, the time evolution of the spec- background and large-scale gal- with the redshifts, but with the dis- trum for the SN axy distributions. The acceleration is tances. Using galaxies to measure dis- 1996bj at z ϭ 0.574 was also stretched attributed to the negative pressure of a tances proved frustrating: the stars that out by the redshift (34). Goldhaber et al. smoothly pervasive component of the (32) examined the effect of time dilation make up galaxies fade as galaxies age, universe: the (4). One by using a large set of high-z supernovae but galaxies accrete more stars, and it possibility is that the dark energy is the and found results in complete accord was too hard to tell whether distant, looked at an- with the expectations of real cosmic ex- young galaxies were intrinsically brighter other way: as a vacuum energy, rather pansion, not photon fatigue. A second or dimmer than their counterparts than as a term in Einstein’s prediction of the expansion idea is that nearby. equations. If this picture is really right, the surface brightness of a galaxy should decrease as (1 ϩ z)4. This ‘‘Tolman dim- ming’’ has finally been observed by Lu- bin and Sandage (35). The idea of tired light has now been put to rest. Einstein’s idea of a , suspended between gravity pulling in- ward and the cosmological constant making the universe expand, was ruled out by Hubble’s data. Legend has it that Einstein, much later, referred to the cosmological constant as his ‘‘greatest blunder’’ (36). In 1947, Einstein wrote, ‘‘Since I introduced this term, I had al- ways a bad conscience. ...Iamunable to believe that such an ugly thing is ac- tually realized in nature’’ (37). In their 1932 farewell to the cosmological con- stant (also published in PNAS), Einstein and de Sitter were more measured: ‘‘an increase in the precision of the data de- Fig. 6. Deviations in the Hubble diagram. Each point in this plot shows the difference at each redshift rived from observations will enable us in between the measured apparent brightness and the expected location in the Hubble diagram in a universe the future to fix its sign and to deter- that is expanding without any acceleration or deceleration. The blue points correspond to median values in eight redshift bins. The upward bulge at z Ϸ 0.5 is the signature of cosmic acceleration. The hint of a mine its value’’ (38). turnover in the data at the highest redshifts, near z ϭ 1, suggests that we may be seeing past the era of The cosmological constant was ban- acceleration driven by dark energy back to the era of deceleration dominated by dark matter. From top ished by Einstein’s curse from serious to bottom, the plotted lines correspond to the favored solution, with 30% dark matter and 70% dark cosmological discussion from 1932 to energy, the observed amount of dark matter (30%) but no dark energy, and a universe with 100% dark about 1995. The observational program matter (from ref. 18).

12 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.2536799100 Kirshner Downloaded by guest on September 28, 2021 then constructing a precise Hubble dia- scope by shuttle astronauts a year ago seven, reference to de Sitter’s strange gram in the era where the acceleration (42, 43). If the dark energy picture is kinematic model, and was not enough has its onset,§ and pushing the Hubble right, we should expect to see back to convince Hubble himself of the real- diagram for type Ia supernovae to red- past the era of acceleration, to an ear- ity of cosmic expansion, but that article shifts beyond 1 will help to pin down lier era of deceleration when dark mat- in PNAS pointed the way to under- the nature of the dark energy. ter ruled the of the universe standing the history of the universe, This work is already underway. Fit- as hinted at by Fig. 6. Hubble hoped to and the continuing search among the tingly, this extension of Hubble’s work understand cosmic expansion by seeing ‘‘ghostly errors of measurement’’ has is being carried out using the Hubble the higher-order terms that lay beyond led to a deeply surprising synthesis of Space Telescope as well as ground- the linear expansion of the nearby dark matter and dark energy. based observatories. One very high sample; we are now looking deep into redshift supernova was discovered in I am very grateful to John Huchra, Saurabh the past at the limits of today’s tech- Jha, , Tom Matheson, and 1996 (41) at redshift 1.7, well over nology to observe directly these John Tonry for help with the text and illus- halfway back to the Big Bang, and changes in cosmic expansion. As trations and to the entire High-Z Super- many more will be found with the new Hubble said in The Realm of the Nebu- nova Search Team for their hard work in Advanced Camera for Surveys that was lae, ‘‘We measure shadows, and we building on Hubble’s foundation. Work on installed on the Hubble Space Tele- search among ghostly errors of mea- supernovae and the Hubble constant at surement for landmarks that are Harvard University is supported by Na- tional Science Foundation Grants AST scarcely more substantial. The search § 02-05808 and AST 02-06329 and by the Na- Kirshner, R. P., Aguilera, C., Barris, B., Becker, A., Challis, P., will continue’’ (44). Hubble’s article Chornock, R., Clocchiatti, A., Covarrubias, R., Filippenko, tional Aeronautics and Space Administra- A. V., Garnavich, P. M., et al. (2003) Bull. Am. Astron. Soc. had velocities from Trumpler without tion through Space Telescope Science Insti- 202, 2308 (abstr.). citation, distances wrong by a factor of tute Grants GO-08641 and GO-09118.

1. Hubble, E. P. (1929) Proc. Natl. Acad. Sci. USA 15, 15. Grego, L., Carlstrom, J. E., Reese, E. D., Holder, 31. Hoeflich, P., Khokhlov, A., Wheeler, J. C., Phil- 168–173. G. P., Holzapfel, W. L., Joy, M. K., Mohr, J. J. & lips, M. M., Suntzeff, N. B. & Hamuy, M. (1996) 2. Bennett, C. L., Halpern, M., Hinshaw, G., Jarosik, Patel, S. (2001) Astrophys. J. 552, 2–14. Astrophys. J. Lett. 472, L81–L84. N., Kogut, A., Limon, M., Meyer, S. S., Page, L., 16. Schmidt, B. P., Kirshner, R. P., Eastman, R. G., 32. Goldhaber, G., Groom, D. E., Kim, A., Aldering, Spergel, D. N., Tucker, G. S., et al. (2003) Astro- Phillips, M. M., Suntzeff, N. B., Hamuy, M., G., Astier, P., Conley, A., Deustua, S. E., Ellis, R., phys. J. Suppl. 148, 1–27. Aviles, R. & Maza, J. (1994) Astrophys. J. 432, Fabbro, S., Fruchter, A. S., et al. (2001) Astrophys. 3. Burles, S., Nollett, K. M. & Turner, M. S. (2001) 42–48. J. 558, 359–368. Astrophys. J. 552, L1–L5. 17. Krauss, L. M. & Chaboyer, B. (2002) Science 299, 33. Leibundgut, B., Schommer, R., Phillips, M., Riess, 4. Kirshner, R. P. (2003) Science 300, 1914–1918. 65–68. A., Schmidt, B., Spyromilio, J., Walsh, J., Suntzeff, 5. Kirshner, R. P. (2002) The Extravagant Universe: 18. Tonry, J. L., Schmidt, B. P., Barris, B., Candia, P., N., Hamuy, M., Maza, J., et al. (1996) Astrophys. J. Exploding Stars, Dark Energy, and the Accelerating Challis, P., Clocchiatti, A., Coil, A. L., Filippenko, Lett. 466, L21–L24. Cosmos (Princeton Univ. Press, Princeton). A. V., Garnavich, P., Hogan, C., et al. (2003) 34. Riess, A. G., Filippenko, A. V., Leonard, D. C., 6. Schmidt, B. P., Suntzeff, N. B., Phillips, M. M., Astrophys. J. 593, 1–24. Schmidt, B. P., Suntzeff, N., Phillips, M. M., Schommer, R. A., Clocchiatti, A., Kirshner, R. P., 19. Phillips, M. M. (1993) Astrophys. J. Lett. 413, Schommer, R., Clocchiatti, A., Kirshner, R. P., Garnavich, P., Challis, P., Leibundgut, B., Spyro- L105–L108. Garnavich, P., et al. (1997) Astron. J. 114, 722–729. milio, J., et al. (1998) Astrophys. J. 507, 46–63. 20. Riess, A. G., Press, W. H. & Kirshner, R. P. (1996) 35. Lubin, L. M. & Sandage, A. (2001) Astron. J. 122, 7. Riess, A. G., Filippenko, A. V., Challis, P., Cloc- Astrophys. J. 473, 88–109. 1084–1103. Science 246, My chiatti, A., Diercks, A., Garnavich, P. M., Gilli- 21. Geller, M. J. & Huchra, J. P. (1989) 36. Gamow, G. (1970) (Viking, New 897–900. York). land, R. L., Hogan, C. J., Jha, S. & Kirshner, R. P. 22. Shectman, S. A., Landy, S. D., Oemler, A., Tucker, 37. Kragh, H. (1996) Cosmology and Controversy (1998) Astron. J. 116, 1009–1038. D. L., Lin, H., Kirshner, R. P. & Schechter, P. L. (Princeton Univ. Press, Princeton, NJ), p. 54. 8. Perlmutter, S., Aldering, G., Goldhaber, G., (1996) Astrophys. J. 470, 172–188. 38. Einstein, A. & de Sitter, W. (1932) Proc. Natl. Knop, R. A., Nugent, P., Castro, P. G., Deustua, 23. Percival, W. J., Baugh, C. M., Bland-Hawthorn, J., Acad. Sci. USA 18, 213–214. S., Fabbro, S., Goobar, A., Groom, D. E., et al. Bridges, T., Cannon, R., Cole, S., Colless, M., 39. Norgaard-Nielsen, H. U., Hansen, L., Jorgensen, (1999) Astrophys. J. 517, pp. 565–586. Collins, C., Couch, W., Dalton, G., et al. (2001) H. E., Aragon S., Alfonso, E. & Richard, S. (1989) 9. Garnavich, P. M., Jha, S., Challis, P., Clocchiatti, Mon. Not. R. Astron. Soc. 327, 1297–1306. Nature 339, 523–524. A., Diercks, A., Filippenko, A. V., Gilliland, R. L., 24. Blanton, M. R., Hogg, D. W., Bahcall, N. A., 40. Perlmutter, S., Gabi, S., Goldhaber, G., Goobar, Hogan, C. J., Kirshner, R. P., Leibundgut, B., et al. Brinkmann, J., Britton, M., Connolly, A. J., A., Groom, D. E., Hook, I. M., Kim, A. G., Kim, (1998) Astrophys. J. 509, 74–79. Csabai, I., Fukugita, M., Loveday, J., Meiksin, A., M. Y., Lee, J. C., Pain, R., et al. (1997) Astrophys. 10. Freedman, W. L., Madore, B. F., Gibson, B. K., et al. (2003) Astrophys. J. 592, 819–838. J. 483, 565–581. Ferrarese, L., Kelson, D. D., Sakai, S., Mould, 25. Sharov, A. S. & Novikov, I. D. (1993) Edwin 41. Riess, A. G., Nugent, P. E., Gilliland, R. L., J. R., Kennicutt, R. C., Jr., Ford, H. C., Graham, Hubble: The Discoverer of the Big Bang Universe Schmidt, B. P., Tonry, J., Dickinson, M., Thomp- J. A., et al. (2001) Astrophys. J. 553, 47–72. (Cambridge Univ. Press, Cambridge, U.K.), p. 67. son, R. I., Budava´ri, T., Casertano, S., Evans, A. S., 11. Saha, A., Sandage, A., Labhardt, L., Tammann, 26. Zwicky, F. (1929) Proc. Natl. Acad. Sci. USA 15, et al. (2001) Astrophys. J., 560, 49–71. G. A., Macchetto, F. D. & Panagia, N. (1997) 773–779. 42. Blakeslee, J. P., Tsvetanov, Z. I., Riess, A. G., Astrophys. J. 486, 1–20. 27. Wilson, O. (1939) Astrophys. J. 90, 634–636. Ford, H. C., Illingworth, G. D., Magee, D., Tonry, 12. Witt, H. J., Mao, S. & Keeton, C. R. (2000) 28. Rust, B. (1974) Ph.D. thesis (Univ. of Texas, J. L., Benı´tez, N., Clampin, M., Hartig, G. F., et al. Astrophys. J. 544, 98–103. Austin). (2003) Astrophys. J. 589, 693–703. 13. Fassnacht, C. D., Xanthopoulos, E., Koopmans, 29. Jha, S. (2002) Ph.D. thesis (Harvard Univ., Cam- 43. Riess, A. G., Strolger, L.-G., Tonry, J., Tsvetanov, L. V. E. & Rusin, D. (2002) Astrophys. J. 581, bridge, MA). Z., Casertano, S., Ferguson, H. C., Mobasher, B., 823–835. 30. Hamuy, M., Phillips, M. M., Suntzeff, N. B., Challis, P., Panagia, N., Filippenko, A. V., et al. 14. Reese, E. D., Carlstrom, J. E., Marshall, J., Mohr, Schommer, R. A., Maza, J., Smith, R. C., Lira, P. (2003) Astrophys. J. Lett., in press. J. J., Grego, L. & Holzapfel, W. L. (2002) Astro- & Aviles, R. (1996) Astron. J. 112, 2438– 44. Hubble, E. P. (1982) The Realm of the Nebulae phys. J. 581, 53–85. 2447. (Yale Univ. Press, New Haven, CT), p. 202.

Kirshner PNAS ͉ January 6, 2004 ͉ vol. 101 ͉ no. 1 ͉ 13 Downloaded by guest on September 28, 2021