The HR Diagram of Globular Clusters:

Theorist’ view(s)
F. D’Antona
arXiv:astro-ph/9910312v1 18 Oct 1999

Osservatorio Astronomico di Roma, Italy

Abstract: I list the characteristic features of Globular Cluster (GC) HR diagrams which provide a
complete test of the stellar evolution of low mass stars: morphologies describing the different evolu-
tionary phases, number ratios and luminosity functions, which add quantitative information. Then I
explore the stage of todays’ understanding of the classical distance scale indicators (providing warn-
ings against model construction which, underneath, already is based on a choice of distance), and
compare them to the “new” indicators, such as the white dwarfs and the first and second kink of the
low main sequence. The classical and new distance indicators are still subject to uncertainties due to
tiny details of the theory, which are all of the same order of magnitude, ∼ 0.25mag and the absolute
ages of GCs can not be constrained to better than 10 – 16 Gyr. However, most of recent theoretical
and observational results (including both classical and new indicators) point more towards the lower
range of ages (10-12) than to the upper range (15-16).

1 Introduction
It is now close to forty years that Globular Clusters (GCs) HR diagrams are used to derive
the age of the oldest stars of the Galaxy, but today they provide a very complete test of the
predictions of stellar evolution of low mass stars. This is easily recognized by looking at the
composite HR diagram of the GC NGC 6397 shown in figure 1, and illustrated in figure 2 by
recent stellar models. The diagram shows the core H–burning phases (main sequence –MS–
and turnoff –TO– regions, the H–shell burning of the evolving mass (M ≃ 0.8M⊙ ) including
the red giant branch –RGB–, the Helium core burning phase (horizontal branch –HB–) and the
double shell burning phase (asymptotic giant branch –AGB–). The Planetary Nebula phase
is too short (only two planetaries are known in the galactic GC system) but the white dwarf
–WD– cooling is well represented. The MS ends at the top with the evolving stars, and at the
bottom with the lowest masses which are able to ignite hydrogen. The characteristic shape of
the MS well below the turnoff displays characteristic details of the physics of the atmospheres
and interiors of low and very low mass stars (see, e.g. D’Antona 1995): it changes slope twice,
at a first, more luminous “kink” (FK) at M ∼ 0.5M⊙ , and at a second kink (SK), very close to
the end of hydrogen burning structures. So the HR diagrams provide morphological information
and reference luminosity indicators (TO, HB, WDs) which give constraints on the cluster stars
evolution. In addition the number ratios (luminosity functions (LFs), “clumps”, gaps, ratio of
HB to RGB number stars) add valuable quantitative information to the morphology.
 In particular, the LF of the turnoff and giant branch depends mostly on the age of the
system, while the mass function affects its unevolved part. The LF below the TO presents a
characteristic broad maximum, due to the functional form of the mass – luminosity relation,
whose peak becomes dimmer when the metal content increase (for a full description see, e.g.,
Silvestri et al. 1998).
In my opinion, the two fundamental questions which we would like to address in a meeting
on the comparison between GCs and halo field stars are the following:

1. is our knowledge sufficient to constrain the GC ages at a level at which they can be
interesting as indicators of the age of the Universe?

2. are GC stars identical to their halo counterparts of similar metallicity?

The answers to these questions are simple and a bit unconfortable:

• Details of the theory determine the precise absolute ages of GCs: till now it is difficult to
constrain the ages to better than from 10 to 16Gyr (but see later);

• The distance scale of GCs is necessary to know their ages. Its calibration necessarily
relies on the hypothesis that GC and halo stars are strict relatives, unless we put all our
faith on theoretical models only.

In the following I describe the ways in which we can try to determine the distance scale of
GCs and the hidden dangers in the playing of the “isochrone fitting computer game”, mainly
those related to the use of the giant branch location. I will finally show how the new distance
indicators emerging in the latest years (WDs, FK and SK of the low MS) are consistent with
the traditional distance scales based on the fit of ground based photometry and on the HB
models.

2 The ages paradigm

The distance scale of GCs is the main key to their age determination. There are many “tradi-
tional” methods to derive this scale: they can be reduced to the following list:

• Approach based on distance indicators:

– Fitting of MS to local sample of subdwarfs;

– Fitting of the GCs RR Lyrae to RR Lyrae in the Magellanic Clouds, distance of
LMC calibrated through the Cepheids;
– Fitting of HB (or RR Lyrae) to local halo HB or RR Lyrae.

• Purely theoretical approach:

– Fitting of HB (or RR Lyrae) luminosity to HB theoretical models;

– Fitting of observed MS to theoretical MS (this implies a match of the models colors);
– Fitting of the morphology of the HR diagram (δ(B − V ) type methods)

The recent, mainly HST based, observations which allow to reach dimmer and dimmer
luminosities have added new ways to determine the distance, or at least to check it, which will
be examined in sections 5, 6 and 7, namely:
 • fitting of the location of the low MS, with attention to the location of the FK;

• fitting of the MS region following the SK with the local M subdwarfs sample;

• fitting of the WD sequence to models or disk counterparts.

I will not discuss the approach based on the classical distance indicators, which has known a
renewed interest in these years, thanks to the impact of the results from the Hipparcos satellite.
The fitting to the local subdwarfs is discussed in Reid 1997, Gratton et al. 1997, Pont et al.
1998 and Chaboyer et al. 1998. The RR Lyrae calibration after Hipparcos data was first
rediscussed by Feast and Catchpole (1997). Notice that, while most Hipparcos results imply
a more or less stringent confirmation of the so called “long” distance scale for GCs, the local
RR Lyrae and HB stars give a much shorter scale (Fernley et al. 1998) consistent with the
previous statistical parallax determination by Layden et al. 1996 –but see the recent approach
by Groenewegen and Salaris 1999.
I will mostly concentrate on the theoretical approach. Actually, none of the theoretical
methods has been ever used independently from the others, but a sort of “consistency” between
the different aspects of the problems has generally been looked for, including the observational
distance indicators. For instance, fitting the morphology (that is, the relative position of RGB
and TO) as an absolute method for the age determination has never been taken seriously (see
later) but the consistency of the whole HR diagram locations (as shown in figure 1b) has been
more or less unconsciously taken as selfevidence of an evolutionary scheme –and thus of a given
range of ages– and especially in some comparisons with observations we have seen mention
of “spectacular fit”, or “location and shape matched superbly by isochrones” while the truth
hidden below is very different.
In the following I will base part of my discussion on a personal interpretation of the events
which in recent years led to a revision of the average age of GCs. A look at the literature in fact
shows that the ages of GCs quoted before or up to 1996-1997 (pre-Hipparcos) are in the range
13-18Gyr, while the most quoted ages of the years 1998-1999 are 10-14Gyr (post-Hipparcos)1 .
Vandenberg et al. 1996 in fact give an age of 15.8 ± 2Gyr to M92, noting that “ages below
12 or above 20Gyr appear highly unlikely”, and Chaboyer et al. (1996) give an average age of
14.6 ± 1.7Gyr to the galactic GCs, putting a 95% lower bound at 12.1Gyr. The turning point
seems to have been the results from Hipparcos satellite, which on the one hand made the metal
poor subdwarf sequence more luminous (by no more than 0.1mag, the effect being even lower
according to some researchers) and on the other hand contributed to raise the zero point of
Cepheids’ luminosity, leading thus to confirm a larger luminosity of the RR Lyrae in the LMC.
However, the Hipparcos results alone do not justify the global shift of the average age of GCs,
which amounts to ∼ 4Gyr (from 16 to 12Gyr). In my opinion, Hipparcos has simply given
more weight to the the “long” distance scale of GCs, which already had some enphasis in the
observational literature (Sandage 1993, Walker 1992).
What has been happening is schematized in a very naif way in figure 3: the HB (or RR
Lyrae) luminosity has been increased in recent models due to the sum of two small effects (a
slight increase in the core mass at the helium flash -by about 0.01M⊙ - and a slight increase
due to the improvement in the equation of state (EoS)). This has been the most important
update in the models, and has shifted the ages to at most a couple of Gyr smaller with respect
1
Two notable exceptions are the work by Salaris et al. (1997) and the three papers by our group (Mazzitelli
et al. 1995, D’Antona et al. 1997, Caloi et al. 1997) which published post–Hipparcos ages in the pre–Hipparcos
years
to previous HB theoretical luminosity2 .
At the same time, the TO luminosity corresponding to a given age has been slighlty de-
creasing. The sum of these subtle effects is a good ∼ 0.27mag of difference in the absolute TO
location at a given age based on old or new models, for a given ∆V from the HB to the TO, and
thus the net effect is a reduction of 4Gyr in the age. The interpretation which figure 3 gives to
the age decrease is not unique: other motivations for a more or less substantial decrease in the
age are found in the recent literature. Pont et al. (1998), who do not revise the HB luminosity,
remark however a difference in the scale of the V bolometric corrections (BCV ) between the
models atmospheres employed by Bergbush and Vandenberg 1992, and the most recent scale
both by Kurucz and by Bell, amounting to ∼0.1mag, and thus leading to a decrease in the age
by ∼ 1.5Gyr.
It is evident that the HB luminosity of the present models, which in the end represent the
most direct classical distance indicator, is in itself still uncertain, unless we believe that we can
trust our models at the level of 0.01M⊙ for the determination of the helium core mass at flash,
and that we know perfectly all the other pieces of input physics. At least, discussion is still open
on the helium core flash masses. Notice also that, e.g., the most recent HB models (Caloi et al.
1997 versus Cassisi et al. 1998) do not agree on the LHB at intermediate metallicity (for [M/H]
from ∼ –1.5 to ∼ −1), and it is unclear why. The problem of GC ages is linked to minute details
of the input physics, and we can not exclude an uncertainty of ∼ 0.25mag in the theoretical
determination of the HB luminosity. So the real uncertainty on the age determination from the
HB is still of ±several Gyr.
However, why the paradigm of the 15-16Gyr age was so difficult to be abandoned? In my
opinion a part of the answer is the following: in the course of many years, the whole theoretical
construction of the GC HR diagram had been adjusted to be consistent with about that age, so
that is resulted very difficult to make drastic changes to that view. This interpretation becomes
more clear by examining the relative location of TO and RGB in stellar models.
The TO color location is discussed in section 3. The input physics may affect it at a
level of ∼ 0.05mag. On the other hand, the RGB colors heavily depend on the treatment of
convection. By changing the ratio mixing length to pressure scale heigth (α = l/Hp ) in the MLT
formulation, the color location of the RGB may vary by tenths of magnitude (e.g. Vandenberg
1983). Although everybody knew that the α choice was ‘ad hoc’, the “old” distance scale had
this interesting outcome: by chance it had the additional bonus of giving a very good fit of the
MS and RGB locations, if the models employed the same α parameter which fitted the solar
radius at the solar age (solar calibration of the mixing length). In addition, the solar calibration
was also in good agreement with the location of the best known subdwarf Groombridge 1830.
It was perhaps necessary to add some very small adjustment of colors, but the reproduction of
the GC morphologies was indeed very good. The best example of this procedure is given by
Bergbush and Vandenberg 1992: they show how they calibrate their color-Teff relations (based
on quite good model atmospheres) to provide a “consistent” picture for all metallicities. They
also explicitly state that the transformations they adopt are OK for their own models, and
that different adjustments might be required by other models. In spite of being very careful,
the procedure adopted in Bergbush and Vandenberg 1992 (but also by others) implicitly hides
both the choice of the distance scale (and thus the resulting 15Gyr or so) and the choice of the
convection model, as I will now clarify. The fortuitous coincidence between the RG location
in population II models with a solar calibrated α led researchers to postpone the problem of
2
Some confusion was however present in the literature up to 1997 as to the absolute visual magnitudes
corresponding to the log L/L⊙ of the models (see e.g. the display in the lower panel of figure 1 from Caloi et al.
1997), so that the modification induced in HB models might appear more drastic in some authors’ comparisons.
a better understanding of superadiabatic convection3 . After 1997 the -even small- change of
distance scale implied by the new HB models and by the Hipparcos subdwarfs re–calibration
did not allow any longer to forget the problem: the same theoretical RGB, for a smaller age,
provides a larger δ(B − V ) and the theoretical RGs were too red. Thus the “new” ages required
a change either in the convection modelling, or a different tuning of the correlations colors-Teff ,
or both. The situation is schematically shown in figure 4, in which I use an extreme difference in
the distance modulus (0.25mag) to clarify the problem. Suppose that color magnitude diagram
was well fit by an isochrone (open squares in figure 4. An update of the distance modulus
to 0.25mag longer (and the adjustment of the color by 0.06mag, in the range of TO color
uncertainties) provides now an age ∼ 6Gyr younger, but it does not allow to fit of the RGB. A
good fit requires a bluer RGB.
The modellers have tried to solve the problem of the discrepancy between the δ(B − V ) and
the new distance scale in the following ways:

1. they have increased α to obtain again the fit. On theoretical grounds, there is no scientific
basis in the assumption that the α in different stars should be the same as in the solar
model, so why not? This solution is adopted e.g. by Brocato et al. (1998) who discuss at
length the effect described here for the case of the GC M68) and by Cassisi et al. 1998;

2. some researchers have considered again models with solar α, but have chosen the color
Teff relation in an appropriate way to reproduce the giant colors (it is generally possible to
find good justifications for this choice also). This is the approach by Salaris and Weiss
1997, 1998: they adopt Buser and Kurucz (1978) colors for the giants, which are bluer by
several hundreths of magnitude than the more recent ATLAS9 updated colors (see e.g.
the comparison in figure 1 of Cassisi et al. 1999). In this way, the δ(B − V ) between the
TO and the RGB results smaller and can fit GCs shapes with the new distance scale.

3. there are a few attempts to try different convection models, which will generally not allow
a “perfect fit”.

If one adopts the solutions 1) or 2), it is important to remember that the shape of the HR
diagram loses any predictive power, as it has been fit already assuming a distance scale: just as
the “spectacular fits” of a few years ago produced a 15Gyr answer, present day fits will produce
a 10-12Gyr answer, but the quality of the fit has nothing to do with the truth of the answer.
A better way of posing the problem, when an observer adopts a given set of tracks to infer the
age of a new stellar system, would be to say that the system shows the same age of -or that its
age differs from- the GC on which the track set has been more or less explicitly calibrated. It is
not clear to me that even relative ages of GCs of different metallicities can be inferred from the
δ(B − V ) or δ(V − I) method, when we use a given MLT prescription which already is tested
on the HR diagrams to fit a given distance scale.
The 3rd solution is less misleading, and it could in the end allow progress in the field, but
it requires lots of work and maybe frustrating, as it produces results not always in “perfect
agreement” with observations4 . A few such attempts to overcome the MLT are today available:
3
This “canonical” assumption was abandoned only in the Mazzitelli et al. 1995 paper, treating convection
according to the Canuto and Mazzitelli (1991) model, and in Chieffi et al. (1995), who propose a calibration of
α as a function of the cluster metallicity. This latter paper puts clearly into evidence that no predictions can
be made on the RGB location on the basis of MLT models.
4
One of the reasons why we could get the hint of a decrease in the GC ages two years before Hipparcos
(Mazzitelli et al. 1995) was that we used a convection models by which it was not possible to get a fine tuning
of the RG location.
 1. the FST (or Full Spectrum Turbulence) models, based on the Canuto and Mazzitelli
(1991) formulation, whose fluxes are in good agreement with experimental data, and are
computed using modern closures of the Navier Stokes equations, and in which the scale
length is assumed to be the distance from the convective boundary. Models have been
computed by Mazzitelli et al. 1995, D’Antona and Mazzitelli 1997, Silvestri et al. 1998.
This formulation of convection gives a different flavour to the HR diagram shape and it is
less tunable than the MLT, an advantage in terms of predictive power, but a real failure
if we want to obtain perfect fits. However, the Silvestri et al. (1998) models, which differ
from the previous of our group mainly for the updated choice of color-Teff transformations
(Castelli 19985 versus Kurucz 1993),also provide a reasonable fit of the RGB as shown in
figures 2 and 6 –but notice also the discrepancy in the case of M30 in figure 5.

2. Freytag and Salaris (1999) have calibrated the MLT α by RHD models based on grids of
2D hydrodinamic simulations by Ludwig et al. 1999. Although numerical simulations are
able to take into account only a relatively small number of eddies for a realistic description
of turbulence, the Freytag and Salaris approach is an interesting novelty for this field.

3. an incongruence of FST models and of models not adopting a plain MLT description is
that in any case they use till now grey boundary conditions, and the colors are obtained
through transformations based on MLT model atmospheres. Models including, e.g., FST
model atmospheres should be built up to get selfconsistent colors (Kupka, Schmidt and
D’Antona 1999);

3 The TO and upper MS location

The location of the TO is affected by many uncertainties in the input physics, although not
at the level of the RGB. If we wish to use the theoretical MS colors to determine an age, we
must shift vertically the cluster HR diagram until it is superimposed to the MS of the observed
metallicity, and then we determine the age from the TO luminosity. The MS is very steep in the
TO region: a simple shift in color of the theoretical MS by +0.02mag implies a determination
of age smaller by 2Gyr, not to talk about the possible uncertainty in the reddening. The main
inputs affecting the MS and TO location are the following:

1. the convection description affects both the TO color and its shape (see the comparison
between the MLT based description and the FST models in Mazzitelli et al. 1995 and
D’Antona et al. 1997);

2. The helium gravitational and thermal settlings (diffusion) affect both the TO color and
the age. A number of models are available, starting from Proffitt and Michaud 1991, up
to D’Antona et al. 1997, Straniero et al. 1997 and Cassisi et al. 1998);

3. the color - Teff relation is affected by the convection treatment in the atmosphere (cf.
Kurucz 1993 versus Castelli 1998 models).

Everything included, the absolute determination of the TO and upper MS colors is uncertain
by ∼ 0.05mag, so that it is better not to rely on colors for age determination.
I add a few words about the possible effect of helium diffusion. It is today well settled that it
is necessary to include the treatment of microscopic helium diffusion to account for some details
5
see Kurucz website http://cfaku5.harvard.edu
of the seismic Sun (Bahcall and Pinsonneault 1995, Basu et al. 1996), but the evaluation of
the diffusion coefficients is difficult, and its application to models not always well clear in the
researchers description. The diffusion affects both the TO color and the age: age reductions
from 5-10% to 20% are found, and the TO color may be affected up to 0.1mag in some models.
An important warning first issued by Deliyannis and Demarque (1991) must be kept in mind:
diffusion affects lithium nearly in the same way as helium, thus: “the properties of the Spite
plateau in population II severely restrict the amount of diffusion induced curvature that can
be tolerated in a lithium isochrone”. In other words, the effect of “too much” diffusion would
appear in a smaller lithium abundance for the hotter population II stars, a fact which is not
verified in the halo subdwarfs, which show a remarkably flat lithium abundance versus Teff (the
Spite and Spite 1982 plateau)6 . Chaboyer et al. 1992 show that an age reduction up to 3Gyr
(15%) is in principle possible for GCs when diffusion is included in the computation, but they
find that the lithium isochrones imply a maximum age reduction by 1Gyr (∼ 7%).
In conclusion, also the theoretical TO – upper MS color location is affected by uncertainties
by which the absolute age determination, again, can not be known to better than ± several
billion years.

4 Location of the lower MS

The “double kink” shape of the low MS of GCs is due to the influence of the interior physics on
the structure of low mass stars. The appearence of the FK is attributed mainly to the to the
lowering the adiabatic gradient when the H2 dissociation begins to be present in envelope (below
∼ 5000K). The SK is associated with the reaching of degeneracy in the interior (D’Antona and
Mazzitelli 1996). The shape of the low MS can be a powerful tool, first to constrain the models,
and then to constrain the GC parameters.
As first shown by Baraffe et al. 1995, when the formation of molecules begin to be important
< 5000K) the grey atmospheric integration fails to give a good
in the stellar atmosphere (at Teff ∼
description for the boundary conditions: in summary, it underestimates the opacities and does
not account for the opacity distribution with wavelength. The net effect is that the grey
integration provides much larger pressure and density at the bottom of the atmosphere. In the
interior, the temperature gradient is the adiabatic gradient, so that finally the same central
conditions give a larger Teff . Figure 5 shows in fact that, for the same chemistry, non–grey
models are redder by ∼ 0.06mag with respect to grey models. This also implies that the
metallicity and probably also the element to element ratios, are important to determine the
location of the FK. This is certainly a powerful tool, but also makes the region below the FK
very dependent on the model inputs.
Te EoS adopted for these low mass models is also an important ingredient. It determines
the slope of the region between the two kinks and, together with the atmospheric integration, it
influences the mass luminosity relation, which is the most important input for the interpretation
of the luminosity functions of the MS in terms of mass function. There are still substantial
uncertainties close to the bottom of the main sequence (see Montalban et al. 1999).
Figure 5 summarizes the uncertainties in the low MS location and part of the uncertainties
in the upper MS and TO locations. We see that there is a region between Mv = 6 and 7
at which the uncertainties in color transformations, convection, diffusion, boundary conditions
6
Here again we attribute to GCs stars the same properties of the nearby subdwarfs. Actually, the lithium
behaviour at the TO of GCs might be a bit different than in the field stars. Boesgaard et al. 1998 show that
the M92 TO stars have a larger scatter in lithium than field stars. The Spite’s plateau must still be confirmed
by extensive GC stars observations, which are becoming possible with the new generation telescopes.
seem to play almost no role: This region, then, is the best to be used as distance indicator for
the MS.

5 Consistency of HB and low MS distance indicators

The new HST data which have so much extended our knowledge on the low luminosity part
of the HR diagram put an interesting problem: is there consistency between the “optical”
traditional distance indicators for GCs and the location and shape of the low MS?
We can check this idea by following this procedure: first we can fit the optical data to the
RR Lyrae (or HB) to derive a distance, and check the reddening by controlling the MS location
at Mv ∼ 6 − 7; then we adopt the same reddening and distance modulus for the MS HST data.
This allows us to see if the MS and first kink location are consistently reproduced.
We show in the figures 6 and 7 the check on the low metallicity ([M/H]=-2) clusters M92
and M30, finding excellent consistency. The comparisons are equally good in the HST color
bands (F555 and F814) and in the transformed Johnson–Cousins bands V and I (in fact these
HST bands and the standard magnitudes are only marginally different). Thus, on the one side
the location of the FK provides a check of the distance, on the other side we gain confidence
in the good quality of the low mass models including non grey boundary conditions.
Here again we must admit that this result is not unique: if we assume for M30 a short
distance modulus, say (m − M)0 = 14.5, the general agreement of the sequences would still be
reasonable (and the age would increase to ∼ 16Gyr). Of course our HB models would not fit
the HB luminosities of the cluster, but we have agreed that even a small change in the input
physics may lead to less luminous HBs7 .

6 The distance through the fit to the lowest MS

This approach has been applied by Reid and Gizis (1998) to the only cluster for which the
lowest main sequence, the post-SK region, is known, namely NGC 6397 (figure 1). This part of
the HR diagram is populated by stars which are close to degeneracy in the interior, so they are
close to the minimum mass which can ignite hydrogen. The mass–luminosity relation here is
very steep, that is, the stars have practically all the same mass, and the HR diagram location
follows a constant radius line. In addition, the degeneracy radius is not so much dependent
on the details of the structure, and this locus is very similar for all GC metallicities. Thus,
although the models are not well understood in detail here (Montalban et al. 1999), if we have a
reasonable sample of M subdwarfs with known distances and accurate colors, we can fit the GC
sequence to the nearby sequence and get the distance modulus. The example of the fit by Reid
and Gizis (1998) is very interesting, and provides for NGC 6397 a not unreasonable modulus
(m − M)0 = 12.13 ± 0.15mag, again consistent with the long distance moduli and short ages.
We need a better definition of the lowest MS through a larger sample of M subdwarfs, and
more GC explored down to the post-SK region, to make this distance indicator more useful.

7
Some further indications on the choice of the “best” distance can come from the comparison of the observed
and theoretical luminosity functions: D’Antona (1998) and Silvestri et al. (1998) show that the non monotonic
mass functions derived, e.g. by Piotto et al. 1997, and by others, were mostly due to the use of too short a
distance modulus for the examined clusters.
7 White dwarfs
There are by now three well defined WD sequences for GCs: NGC6752 (Renzini et al. 1996),
NGC6397 (King et al. 1998, see figure 1) and M4 (Richer et al. 1997). For the two latter
clusters, the location of the WD sequence is consistent with the cooling track for M ≃ 0.5M⊙
by Wood (1995) models transformed into the observed magnitudes by means of Bergeron et al.
(1995) model atmospheres (see Richer et al. 1997). The errors on the reddening of the clusters
and on the model observational colors are yet such that we can not quantify better this general
statement. The data of NGC6752 have been compared by Bragaglia et al. with a “standard”
sequence of field WDs spectroscopically determined to be of mass ≃ 0.5M⊙ (from the list of
Bragaglia et al 1995). They obtain a “short” distance modulus, and an age of ∼ 15Gyr (even
assuming a large α–enhancement). This result is then marginally discrepant from the others
we have quoted so far.
On the other hand, from Wood (1995) models we see that ∆Mv /∆Mwd ≃ 2.5. In other
words, a small uncertainty (by 0.1M⊙ ) in the mass determination fo the field WD sample
produces a noticeable difference (by 0.25mag) in the WD sequence location, which results in an
age difference close to 4Gyr. Do we know the spectroscopic masses within 0.1M⊙ ? We know
that the spectroscopic masses of helium atmosphere WDs are highly uncertain, some uncertainty
surely weighs also on the DA type WDs. It is also possible that the field WDs differ from the
GC WDs in other ways: the environment in which they are born is very different and may
imply, e.g., different accretion rates, much larger in the disk than in the GC, which is devoid
of gas and dust. Although the sedimentation of metals is very efficient in these high gravity
stars, it is well possible that some residual effect from accretion affects the stellar radius.

8 Summary
I conclude with the following short summary:
– Morphological fits including the RGB are meaningless in terms of age determination;
– The traditional best theoretical distance scale still mostly relies on HB models, and on
the MS colors at Mv ∼ 6 − 7.
– the small versus large ages only require a difference of ≃ 0.25mag of distance modulus;
– The uncertainties in HB, TO, MS, WDs sequences location are all of the same order of
magnitude: namely: ∼ 0.25mag;
– then we can make no firm choice between 10-12 and 14-16Gyr age;
– however most recent theoretical and observational results, including the new constrains
on distance by the very low luminosity stars, all point towards the smaller range of ages.

Acknowledgements
I warmly thank the organizers of the Colloquium, and especially Arlette Noels, first for inviting
me, and second for holding a perfect scientific and logistic meeting.
I also thank my coworkers and friends Vittoria Caloi and Italo Mazzitelli, together with
Josefina Montalban, Paolo Ventura, and Fabio Silvestri for always useful exchange of ideas and
for the work which has made this review possible. I acknowledge all the useful data and lively
information from F. Allard, A. Cool, I. King, G. Piotto, H. Richer and M. Salaris.
References

Andreuzzi et al. 1999, submitted to A&A

Bahcall, J.N., & Pinsonneault, M. 1995, Rev. Mod. Phys. 67, 781
Baraffe I., Chabrier G., Allard F., Hauschildt P., 1995 ApJ 446, L35
Baraffe I., Chabrier G., Allard F., Hauschildt P., 1997 A&A 327, 1054 (BCAH97)
Bergeron, P. Wesemael, F. & Beauchamp, A. 1995, PASP 107, 1047
Basu, S., Christensen Dalsgaard, J., Schon, J., Thompson, M.J. 1997, ApJ 460, 164
Bergbusch, P.A., & VandenBerg, D.A. 1992, ApJS 81, 163
Boesgaard, A.M. et al. 1998, ApJ 493, 206
Bragaglia, A., Renzini, A., Bergeron, P. 1995, ApJ 443, 735
Buonanno, R., Corsi, C. & Fusi Pecci, F. 1985 A&A 145, 97
Buser, R., Kurucz, R. 1978, A&A 70, 555
Caloi, V., D’Antona, F., & Mazzitelli, I. 1997, A&A 320, 823
Canuto, V.M., & Mazzitelli, I. 1991, ApJ 370, 295
Cassisi, S., Castellani, V., Degl’Innocenti, S., Weiss, A. 1998, A&AS 129, 267
Cassisi, S., Castellani, V., Degl’Innocenti, S., Salaris, M., Weiss, A. 1999, A&AS 134, 103
Castelli, F. 1998, in “Views on distance indicators”, Mem.S.A.It. 69, 165
Chaboyer, B. 1995, ApJ 444, L9
Chaboyer, B., Deliyannis, C.P., Demarque, P., Pinsonneault, M.H., Sarajedini, A. 1992,
ApJ 388, 372
Chaboyer, B. & Kim, Y.–C. 1995, ApJ 454, 767
Chaboyer, B., Demarque, P., Kernan, P.J., Krauss, L.M. 1996, Science 271, 957
Chaboyer, B., Demarque,P., Kernan, P.J., Krauss, L.M. 1998, ApJ 494, 96
Chieffi, A., Straniero, O., Salaris, M. 1995, ApJL 445, L39
Cool, A. M. 1997, “Binary Stars Below the Turnoff in Globular Cluster Color–Magnitude
Diagrams” in Advances in Stellar Evolution, eds. R. T. Rood and A. Renzini (Cam-
bridge: Cambridge U. Press), p. 191
D’Antona F., 1995,in The Bottom of the Main sequence – and beyond, WSO Astrophysics
Symposia (Springer) Ed. C. Tinney p.13
D’Antona F. 1998, in “The Stellar Initial Mass Function”, ed. G.Gilmore & D. Howell
ASP Conference Series, (San Francisco: ASP), vol. 142, p. 157
D’Antona F., Caloi V., Mazzitelli I., 1997, ApJ 477, 519
D’Antona F., Mazzitelli I., 1996, ApJ 456, 329
Deliyannis, C. & Demarque, P. 1991, ApJ 379, 216
Feast, M.W. & Catchpole, R.M. 1987, MNRAS 286, L1
Fernley, J. Barner, T.G., Skillen, I., Hawley, S.L., HAnley, C.J., Evans, D.W., Solano, E.,
Garrido, R. 1998, A&A 330, 515
Freytag, B. & Salaris, M. 1999, ApJL 513, L49
Gratton R.G., Fusi Pecci F., Carretta E., Clementini G., Corsi C.E., Lattanzi M., 1997,
ApJ 491, 749
Groenewegen, M.A.T. & Salaris, M. 1999, A&A 348, L33
Kaluzny, J. 1997, A&AS 122,1
King I.R., Anderson J., Cool A.M., Piotto G., 1998, ApJ 492, L37
Kupka, F., Schmidt, M. & D’Antona, F. 1999, in preparation
Kurucz, R.L. 1991, in: Stellar Atmospheres: Beyond the Classical Models, L. Crivellari,
I. Hubeny, D. G. Hummer eds., NATO ASI Series (Dordrecht: Kluwer), p. 441
Kurucz, R.L. 1993, ATLAS9 Stellar Atmosphere Programs and 2 km/s grid (Kurucz
CD-ROM No 13)
Layden, A.C., Hanson, R.B., Hawley, S.L., Klemola, A.R., Hanley, C.J. 1996, AJ 112,
2110
Ludwig, H.-G., Freytag, B. & Steffen, M. 1999, A&A 346, 111
Mazzitelli, I., D’Antona, F. & Caloi, V. 1995, A&A 302, 382 (MDC)
Montalban, J., D’Antona, F., Mazzitelli, I. 1999, submitted to A&A
Piotto et al. 1990 ApJ 350,662
Piotto G., Cool A. & King I.R., 1997, AJ, 113 1345
Pont, F., Mayor, M., Turon, C. & Vandenberg, D. A. 1998, A&A 329, 87
Proffitt, C.R. & Michaud, G. 1991, ApJ 371, 584
Reid I.N., 1997, AJ 114, 161
Reid I.N., & Gizis, J.E. 1998, AJ 116, 2929
Richer, H. et al. 1997, ApJ 484, 741
Rogers F.J., Swenson F.J., Iglesias C.A., 1996, ApJ 456, 902
Salaris, M., Chieffi, A., and Straniero, O. 1993, ApJ 414, 580
Salaris, M., Degl’Innocenti, S., and Weiss, A. 1997, ApJ 479, 665
Salaris, A., Weiss, A. 1997, A&A 327, 107
Salaris, A., Weiss, A. 1998, A&A 335, 943
Sandage, A., 1993 AJ 106, 703
Silvestri F., Ventura, P., D’Antona F., Mazzitelli I., 1998, ApJ 509, 192
Spite, F. & Spite, M. 1982, A&A 115, 357
Stetson, P.B. & Harris, W.E. 1988, AJ 96, 909
Straniero, O., Chieffi, A., Limongi, M. 1997, ApJ 490, 425
VandenBerg, D.A. 1983, ApJS 51, 29
VandenBerg, D.A. 1992, ApJ 391, 685.
VandenBerg, D.A., & Bell, R.A. 1985, ApJS 58, 561
VandenBerg, D.A., Bolte M., & Stetson, P.B. 1990, AJ 100, 445
VandenBerg, D.A., Stetson, P.B. & Bolte M. 1996, ARAA 34, 461
Walker, A.R. 1992, ApJ 300, L81
Wood, M.A. 1995, in “White Dwarfs”, ed. D. Koester & K.Werner (Berlin:Springer),
p.41
Figure 1: Composite HR diagram for the stars in NGC6397. The upper part is adapted from
Kaluzny 1997, the giants and turnoff data are from Cool 1997, and the main sequence down to
its low end, plus the white dwarfs are from King et al. 1998.

Figure 2: Main evolutionary and structural phases in the HR diagram: WD: 0.5M⊙ CO cooling
track from Wood 1995; MS: from Baraffe et al. 1997 (BCAH97) for [M/H]=-1.5; TO+RGB:
12Gyr isochrone from Silvestri et al. 1998; HB from Caloi et al. 1997; AGB is schematized.

Figure 3: In recent years, very small variations in the HB model luminosity and in the TO
luminosity for fixed age have produced a relatively huge effect on the age determination. The
“OLD” lines represent a typical location in absolute visual magnitude of HB (top) and TO
(bottom) models in old stellar models. A small increase in the HB luminosity (by ∼ 0.07mag)
has been due to the update in the interior EoS of HB models (Rogers et al. 1996), another
increase by ∼ 0.07mag has been due to a small increase in the core mass at the Helium flash
(see Caloi et al. 1997). At the same time, the TO luminosity at a given age has been slightly
decreased by updates in the EoS (again it is an effect of the Rogers et al. 1996 EoS, found
by Chaboyer and Kim 1995 and D’Antona et al. 1997) and another small decrease (again by
∼ 0.07mag can be attributed either to the convection model (a shape effect due to the Canuto
and Mazzitelli 1991 treatment of convection, in Mazzitelli et al. 1995 and D’Antona et al.
1997), and/or to the influence of helium diffusion (gravitational settling). In total, the new
models give a ∆V between TO and HB which is ∼.27mag larger, and therefore a given ∆V
corresponds to an age ∼ 4Gyr lower.

Figure 4: A schematic view of the problem with the RGB location is shown: a good fit with
a distance scale which would provide 16Gyr age for a cluster, is no longer OK if the distance
modulus is increased. Here we show an increase by 0.25mag to enphasize the problem, but the
situation is similar if the shift is smaller. In practice, a more efficient convection, or a different
color – Teff relation for the RGB, is necessary to fit again the RGB, as shown by the dotted
line. In this sense, all set of tracks have their convection treatment plus color-Teff relations
“calibrated” on a distance scale.

Figure 5: In the plane Mv versus V − I the effect of some important physical inputs on the
HR diagram location is shown. The low main sequence location is about 0.06mag redder when
models are computed with non grey boundary conditions (Montalban et al. 1999), with respect
to the grey atmosphere models (Silvestri et al. 1998). Luckily, the color-Teff relations for the
low main sequence are not so dependent on the employed transformations, as shown by the
location of the Silvestri et al. models transformed via the Allard and Hauschildt 1997 NextGen
models (AH) or with Castelli 1998 models. On the contrary, the transformations produce a
small shift of the turnoff color, by ∼ 0.03mag. At the TO, also the treatment of convection
affects the isochrone colors. The less critical models are those of masses ∼ 0.6 − 0.65M⊙ and
Mv ∼ 6 − 7.5, where the non-grey boundary conditions are not yet important, and where the
effect of superadiabaticity and also the effect of diffusion are minimum. This is then the best
region where to require a fit of the theoretical and observational main sequence.
Figure 6: Composite HR diagram for M92. The left part shows the fit of the optical ground
based data in V versus B − V , including the HB (from Buonanno et al. 1985) and the
RGB+TO+MS data (from Stetson and Harris 1988). The right part are the Andreuzzi et
al. HST data shifted at the distance of and dereddened as the optical data. We adopt
E(V − I) = 1.3E(B − V ) based on Allard and Hauschildt model atmospheres. The isochrones
are from Montalban et al. (1999) for the indicated chemistry.

Figure 7: The same comparison is shown for M30. The optical data (from Piotto et al. 1990 for
the MS and TP, and from Buonanno et al. 1985 for the RGB and HB) are fitted by assuming
a distance modulus of 14.7 and E(B-V)=0.04. The same choice provides a good reproduction
of the HST data for the MS including the first kink. The dashed line here and in figure 6 is
the corresponding 12Gyr isochrone from D’Antona et al. 1997 (DCM), which employed Kurucz
(1993) color transformations and grey atmosphere models.
This figure "f1.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f2.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f3.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f4.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f5.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f6.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1
This figure "f7.gif" is available in "gif" format from:

http://arxiv.org/ps/astro-ph/9910312v1