Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/~ambn/265csj.ps Äàòà èçìåíåíèÿ: Tue Apr 15 15:50:37 1997 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:10:03 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 106

Mon. Not. R. Astron. Soc. 000, 000--000 (0000) Printed 25 November 1996 (MN L A T E X style file v1.3)
Timing the eclipse of HD185510 ?
C. Simon Jeffery 1;2 and Theodore Simon 3
1 Armagh Observatory, College Hill, Armagh BT61 9DG, Northern Ireland
2 School of Physics and Astronomy, University of St Andrews, St Andrews, FIFE KY16 9SS, Scotland
3 Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honoloulu, HI 96822, USA
Accepted . Received ; 1996 September 6.
ABSTRACT
HD185510 (=V1379 Aql) is an eclipsing double­lined RS CVn binary containing a K0
III/IV giant and a hot subluminous companion. An IUE ultraviolet light curve has been
obtained through the eclipse of the companion with an average time resolution of one
measurement every 19 minutes. This was sufficient to resolve all four times of contact
and, with data from previous work on this system, to derive a solution for the absolute
dimensions of the system.
The atmospheric parameters of both components have been investigated. From its
out­of­eclipse flux distribution, the K0 star has T eff = 4 500 \Sigma 300 K and appears to
be metal­deficient. However these measurements are sensitive to the fractional spot
coverage at the time photometry was obtained.
The IUE flux distribution, low­resolution Ly­ff profile and a noisy high­resolution
UV spectrum of the hot companion have been analysed to obtain T eff = 31 500 \Sigma 1 500K,
log g = 7:2 \Sigma 0:3 and EB\GammaV = 0:13 \Sigma 0:03. C and Si are deficient by ¸ 1 dex; it is not
possible to deduce other metal abundances from the existing data. The spectroscopic
gravity is higher than indicated by the eclipse geometry, indicating a weakness in the
Ly­ff analysis. Neglecting Ly­ff, an alternative solution of T eff = 31 000 \Sigma 1 500K, log g =
6:5 \Sigma 0:2, and EB\GammaV = 0:1 \Sigma 0:03 is obtained from the orbital solution below. The relative
diameters of the subdwarf and K giant are r s =r p = 0:0058 \Sigma 0:0010.
These data were used with the eclipse geometry in an attempt to obtain the or­
bital inclination. In order to avoid severe contradictions with other diagnostics it was
necessary to introduce a non­negligible eclipse due to the cool star atmosphere. Whilst
available data favour i = 90 \Sigma 5 ffi , uncertainties introduced by the atmospheric eclipse
meant that i ¸ 80 ffi is also possible.
With i = 90 ffi , the masses of the cool and hot stars are 2:27 \Sigma 0:17 and
0:304 \Sigma 0:015M fi respectively. Whilst the high­gravity (Ly­ff) solution indicates the
identification of the hot star as a helium white dwarf, the orbital solution favours an
identification with sdOB stars such as SB 707. A final resolution of the nature of the
hot star is limited partly by data quality but also, to a large extent, by the intrinsic
properties of the system itself.
Key words: binaries: eclipsing, stars: fundamental parameters, stars: individual:
HD185510.
1 INTRODUCTION
V1379 Aql = HD185510 is an eclipsing double­lined RS CVn
binary containing a K0 giant and a subluminous B star. As
such it presents an opportunity to study important astro­
nomical phenomena, including post­main­sequence binary­
star evolution and cool star chromospheric activity, in an
object whose fundamental properties can, in principle, be
? Based on observations with the International Ultraviolet Ex­
plorer Satellite
determined with great precision. In a previous investigation,
Jeffery et al. (1992) used high­resolution ultraviolet spectra
to determine the mass ratio and analyzed the eclipse light
curve to place constraints on the dimensions of the stellar
components. Residual imprecision remained owing to the
rapidity of eclipse ingress which, being most reliably ob­
served in the space ultraviolet, could not be resolved with
the International Ultraviolet Explorer's (IUE) conventional
photometric procedures, and from the absence of precise ob­
servations of both ingress and egress at a single eclipse. This
paper reports on new observations with IUE which fully re­

2 C.S.Jeffery & T.Simon
solve both events, and which allow the orbital inclination
and primary radius to be established.
Meanwhile, Fekel et al. (1993) have made substantial
improvements to the orbit of the primary star from ground­
based measurements. In addition, they noted that Jeffery et
al. (1992) had adopted a temperature for the cool star that
was inconsistent with the spectral type, and that there was
an inconsistency between the radius obtained for the cool
star from the eclipse light curve and that obtained from its
rotational velocity and period. The opportunity has there­
fore been taken to improve the analysis of the photospheric
parameters of both stars.
2 OBSERVATIONS
Ultraviolet observations of HD188510 were obtained during
the ingress and egress of the eclipse which occurred during
September 11--13, 1993. Both cameras of the International
Ultraviolet Explorer (IUE) were used in low resolution. Ex­
posures were obtained with the target located at a number
of positions in the large aperture (LAP) before reading out
an image. Two positions were used for the long­wavelength
(LWP) camera, and four for the short­wavelength (SWP)
camera. This strategy maximised the time resolution of the
observations, yielding an average of one exposure every 19
minutes, with each exposure lasting 10 minutes and the
longest interval between exposures being 37 minutes. The
dead time was required to read and prepare the cameras
and to trim the spacecraft. Table 1 shows the sequence of
observations, including IUE image numbers, UT start times
of individual exposures, corresponding Julian day numbers
and orbital phase calculated according to the ephemeris de­
scribed below.
3 PHOTOMETRIC MEASUREMENTS
Our objective was to define the light curve of HD185510 due
to the hot component alone. Since eclipse is total, and the
cool star contributes negligible light in the UV shortward
of 2700 š A, this light curve can be defined relative to F=1
out of eclipse and F=0 in eclipse. This simplifies the data
reduction procedure, which must use the multiple IUE ex­
posures of HD185510 to obtain accurate photometry. In the
case of the LWP images, the two exposures are well resolved
spatially and can be extracted separately in the usual way.
However, for consistency, we preferred to treat the LWP and
SWP images in the same way. The four exposures in the
SWP images are heavily blended. For example, at 1250 š A,
the FWHM of the SWP point spread function in the cross­
dispersion direction is ¸ 3 resolution elements (pixels), the
average separation of the images is ¸ 2:8 pixels.
The STARLINK data reduction package IUEDR (Gid­
dings et al. 1994) was used to extract the line­by­line
spectrum (LBLS) from the (IUESIPS) photometrically cor­
rected image sampled every
p
2=2 resolution elements. Sec­
tions through the LBLS spectrum were taken in the cross­
dispersion direction at several wavelengths, each section hav­
ing a full­width of 50 š A. Each section is designated S i (x; –),
where x is the geometric cross­dispersion in pixels, – is the
central wavelength of the section, i is the image number, and
S has the units of flux number. What S should represent is
the photometric behaviour of the star in discrete time inter­
vals as a function of wavelength, Fik = F (x; xik ; –), where
k corresponds to the kth exposure in image i, xik corre­
sponds to the offset of the star in the LAP when the image
was exposed and F has the units of flux. This is convolved
with the point­spread­function of the IUE camera, which is
a function of geometric position of the star in the LAP and
of wavelength, Pk = P (x; xik ; –), and added to background
illumination and scattered light B i (x; –), so that
S i (x; –) =
X
k
F (x; xik ; –) \Lambda P (x; xik ; –)
+B i (x; –) (1)
Additional effects must be taken into consideration.
Variations in IUE camera sensitivity across the detector
(X) are not entirely removed in the photometrically cor­
rected images, and there is also evidence for time variability
in the camera sensitivity (T ). The cross­detector variations
X(x;–) can be approximately calibrated where all exposures
on a single image have been obtained out of eclipse. Time­
dependent variations T i (t) can also be approximately cali­
brated where an image contains at least one exposure out of
eclipse. Thus what is actually measured by IUE is
S i =
/ X
k
Fik \Lambda Pk + B i
!
XT i (2)
If Pk ; X;B i and T i , are known, it is straightforward to
determine the required values Fik from S i . In practice, B i is
determined from a least­squares linear fit to the background
either side of the image. Pk is determined from resolved
exposures of HD188510 (e.g. SWP48627, LWP26352). The
background is heavily smoothed, and Pk is renormalized to
give
Pk (j x jAE 0) = 0 and
Z
Pk (x)dx = 1 (3)
In this paper, we assumed Pk similar for all k. The coef­
ficient products \Pi ik = FikXT i ; (k = 1; K); in the linear sum
(2) can be deduced from the observed S i by linear least­
squares if the LAP image offsets (xik ; k = 1; K) are known.
Where possible, \Pi ik and xik were obtained simultaneously
by ü 2 minimization, using the Levenburg­Marquardt pro­
cedure (Press et al. 1989), otherwise average values were
adopted. Finally, the normalization to F = 1 out of eclipse,
and F = 0 in eclipse allowed Xk and T i to be sufficiently de­
fined that Fik during eclipse ingress and egress could be con­
fidently measured. The data proved to give the best results
for – near the peak of the product of hot­star flux distribu­
tion and camera sensitivity functions, namely at – = 1400 š A
(SWP) and – = 2600 š A (LWP). Table 1 shows Fik for all
exposures k in all images i, together with the covariance
in F; oe ik , obtained at these wavelengths and at – = 1800 š A
(SWP). Table 1 also contains, for completeness, correspond­
ing data for the eclipse ingress observed in 1987 and re­
ported by Jeffery et al. (1992). Eclipse ingress is well de­
fined. Eclipse egress immediately precedes a break in the
photometry due to earth occultation, but is otherwise well

Timing the eclipse of HD185510 3
Table 1. Ultraviolet light curve for HD 185510 obtained during the eclipses of October 25, 1987 and September 11--13, 1993
Image Exp Year UT HJD (mid) Phase F 1400 \Sigma F 1800 \Sigma
i k Day (start) \Gamma2440000 F 2600
Eclipse ingress, October 1987
SWP32157 298 06:09:00 7093.7655 \Gamma0.03583 0.993
LWP11947 298 06:43:00 7093.7867 \Gamma0.03481 0.978
SWP32158 298 07:13:00 7093.8099 \Gamma0.03368 0.766
LWP11948 298 07:49:00 7093.8325 \Gamma0.03259 0.046
SWP32159 298 08:22:00 7093.8561 \Gamma0.03145 0.003
SWP32161 298 11:04:00 7093.9703 \Gamma0.02592 \Gamma0.001
LWP11949 298 11:36:00 7093.9901 \Gamma0.02496 \Gamma0.047
SWP32162 298 12:08:00 7094.0196 \Gamma0.02354 \Gamma0.002
Eclipse, September 1993
SWP48609 1 254 21:17:04 9242.3869 \Gamma0.04157 0.992 0.160 1.076 0.065
SWP48609 2 254 21:37:20 9242.4010 \Gamma0.04089 0.991 0.208 1.100 0.085
SWP48609 3 254 21:52:05 9242.4112 \Gamma0.04040 1.014 0.240 0.845 0.194
SWP48609 4 254 22:06:30 9242.4213 \Gamma0.03991 1.020 0.254 0.555 0.154
LWP26339 1 254 22:25:20 9242.4343 \Gamma0.03928 1.030 0.137
LWP26339 2 254 22:47:06 9242.4494 \Gamma0.03855 1.021 0.137
SWP48610 1 254 23:17:20 9242.4704 \Gamma0.03753 1.008 0.170 0.924 0.084
SWP48610 2 254 23:33:20 9242.4816 \Gamma0.03699 1.009 0.191 0.987 0.099
SWP48610 3 254 23:47:53 9242.4917 \Gamma0.03651 0.986 0.180 0.908 0.106
SWP48610 4 255 00:05:26 9242.5038 \Gamma0.03592 0.980 0.169 0.881 0.076
LWP26340 1 255 00:24:03 9242.5168 \Gamma0.03529 0.970 0.135
LWP26340 2 255 00:38:56 9242.5271 \Gamma0.03479 1.013 0.125
SWP48611 1 255 01:07:14 9242.5468 \Gamma0.03384 0.902 0.234 1.032 0.119
SWP48611 2 255 01:23:27 9242.5580 \Gamma0.03329 0.521 0.217 0.820 0.138
SWP48611 3 255 01:35:22 9242.5663 \Gamma0.03289 0.091 0.058 0.116 0.055
SWP48611 4 255 01:50:34 9242.5769 \Gamma0.03238 \Gamma0.007 0.036 0.006 0.039
LWP26341 1 255 02:09:26 9242.5900 \Gamma0.03175 \Gamma0.016 0.081
LWP26341 2 255 02:24:32 9242.6004 \Gamma0.03124 0.003 0.074
SWP48612 1 255 02:51:10 9242.6189 \Gamma0.03035 \Gamma0.002 0.025 0.028 0.018
SWP48612 2 255 03:07:12 9242.6301 \Gamma0.02981 0.007 0.028 0.009 0.021
SWP48612 3 255 03:21:54 9242.6403 \Gamma0.02931 \Gamma0.006 0.031 \Gamma0.011 0.027
SWP48612 4 255 03:39:15 9242.6523 \Gamma0.02873 \Gamma0.003 0.034 \Gamma0.006 0.021
LWP26348 1 256 03:42:46 9243.6549 0.01980 0.001 0.078
LWP26348 2 256 03:58:02 9243.6655 0.02031 0.008 0.070
SWP48621 1 256 04:29:03 9243.6871 0.02135 0.002 0.023 \Gamma0.004 0.019
SWP48621 2 256 04:43:13 9243.6969 0.02183 0.005 0.024 0.023 0.023
SWP48621 3 256 04:57:36 9243.7069 0.02231 0.000 0.028 \Gamma0.011 0.030
SWP48621 4 256 05:12:15 9243.7171 0.02281 \Gamma0.001 0.031 0.006 0.023
LWP26349 1 256 05:29:07 9243.7288 0.02337 0.019 0.071
LWP26349 2 256 05:45:20 9243.7401 0.02392 0.023 0.076
SWP48622 1 256 06:08:32 9243.7562 0.02470 0.006 0.025 0.020 0.022
SWP48622 2 256 06:28:43 9243.7702 0.02538 0.013 0.030 0.014 0.025
SWP48622 3 256 06:44:22 9243.7811 0.02590 \Gamma0.004 0.031 \Gamma0.010 0.031
SWP48622 4 256 06:58:20 9243.7908 0.02637 0.005 0.032 0.011 0.022
LWP26350 1 256 07:35:43 9243.8167 0.02763 0.005 0.077
LWP26350 2 256 07:51:41 9243.8278 0.02816 0.002 0.075
SWP48623 1 256 08:22:15 9243.8490 0.02919 0.006 0.026 0.063 0.035
SWP48623 2 256 08:41:10 9243.8622 0.02983 0.009 0.026 0.043 0.043
SWP48623 3 256 08:57:17 9243.8734 0.03037 \Gamma0.009 0.029 0.012 0.043
SWP48623 4 256 09:21:16 9243.8900 0.03118 0.001 0.035 0.033 0.031
LWP26351 1 256 09:39:27 9243.9026 0.03179 0.011 0.083
LWP26351 2 256 09:56:52 9243.9147 0.03237 0.006 0.077
SWP48624 1 256 10:23:56 9243.9335 0.03328 0.481 0.117 0.520 0.065
SWP48624 2 256 10:46:29 9243.9492 0.03404 0.927 0.181 0.868 0.077
SWP48624 3 256 11:02:59 9243.9606 0.03459 0.943 0.176 0.925 0.104
SWP48627 1 256 14:29:31 9244.1041 0.04154 0.983 0.109 0.950 0.050
LWP26352 1 256 14:42:16 9244.1122 0.04193 0.774 0.115

4 C.S.Jeffery & T.Simon
Figure 1. The normalized ultraviolet light curve (F 1400 š A +
F 2600 š A ) through eclipse obtained with IUE in October 1987 (open
circles) and September 1993 (filled circles).
defined. Light maximum was recovered before the occulta­
tion commenced, as indicated by the post­occultation SWP
photometry. The low value of the final LWP datum is not
understood, but suggests a possible occultation of the sec­
ondary star by a prominence loop extending above the trail­
ing limb of the K star primary.
The light curve obtained at 1400 š A and 2600 š A was com­
pared with one obtained by the same procedure at 1800 š A
and 2600 š A. Ingress was found to occur some 0:004 \Sigma 0:001
days later, and egress some 0:001 \Sigma 0:001 days earlier, al­
though the errors in flux measurements are compatible with
both light curves being identical. Owing to the noisier data
at 1800 š A, the shorter­wavelength data define the ephemeris
presented below.
4 LIGHT CURVE SOLUTION
Figure 1 shows the eclipse light curve defined by the mea­
surements given in Table 1. The sharpness of eclipse ingress
(and egress) provides a precise reference for establishing the
eclipse period using a six­year baseline, whilst this first ac­
curate measurement of the eclipse duration allows the zero­
point for the eclipse ephemeris to be improved. From the
current data and those obtained in 1987, together with the
constraint that P ¸ 20:6 days, we obtain the new eclipse
ephemeris given in Table 2. The period compares favourably
with that obtained from radial­velocity measurements (Fekel
et al. 1993).
In order to determine other elements of the binary sys­
tem, it is necessary to quantify the duration of eclipse (OE ec)
and the gradient of the flux curve during ingress and egress.
Here, OE ec is defined as the interval between F = 0:5 on
ingress and F = 0:5 on egress. The ingress gradient is char­
acterised by the interval (OE in ) between F = (1 \Gamma \Delta) and
F = \Delta. Since, despite the improved time resolution ob­
tained, the light curve is still undersampled, these quantities
have been derived by defining the function
F (t) = 0:5(1 \Gamma sin(ú=2 min(1; max((t \Gamma tec)=t in ); \Gamma1);
Figure 2. The flux distribution of HD185510 during eclipse (IUE
LWP and UBVRIMN[12] photometry) dereddened with EB\GammaV =
0:13, and the Kurucz (1991a) model flux distributions for with
T eff = 4500 K, log g = 2:5, [Fe=H] = \Gamma1:0.
t ! t0
F (t) = 0:5(1 \Gamma sin(ú=2 min(1; max((tec \Gamma t)=t in ); \Gamma1);
t ? t0 (4)
where t0 is the time of mid eclipse, tec = P:OEec=2 and t in =
P:OE in , and t in and tec are found by minimizing the square
residuals relative to the observed fluxes. In this solution,
\Delta = 0:1 was adopted. The values obtained for the critical
phase intervals were OE ec = 0:06663 \Sigma 0:00009 cycles and
OE in = 0:00092 +0:00009
\Gamma0:00029 cycles. y
5 EFFECTIVE TEMPERATURES AND
ANGULAR DIAMETERS
Before a complete solution for the stellar dimensions can
be obtained, it is necessary to determine their relative di­
ameters. The use of a black­body flux distribution (Tbb =
4000K) by Jeffery et al. (1992) to match the IUE LWP and
UBVRI photometry of the cool component in HD185510
was criticized by Fekel et al. (1993) because it conflicts
with the spectral classification (K0 III/IV) which demands
T eff ¸ 4800K. The dilemma is that in order to match the
UBVRI photometry with such a temperature, it is necessary
to introduce some reddening (EB\GammaV ¸ 0:1) to the system,
in violation of the result EB\GammaV = 0:0 \Sigma 0:02 obtained for the
hot component by Jeffery et al. (1992). We have therefore
attempted a more detailed analysis of the surface parame­
ters of both stars.
y The equivalent result for the flux curve obtained at 1800 š A
is given by OE ec = 0:06641 \Sigma 0:00013 cycles and OE in =
0:00047 +0:00007
\Gamma0:00009 cycles. The egress gradient characteristic (OE eg )
is much less well constrained at OE eg = 0:00130 +0:00040
\Gamma0:00103 cycles
(OE eg = 0:00149 +0:00029
\Gamma0:00126 cycles at 1800 š A).

Timing the eclipse of HD185510 5
Table 2. Orbital Elements and Stellar Parameters for HD185510, given an adopted inclination i = 90 ffi .
Quantity p s units source
Velocity solution
P 20.66187 \Sigma0.00058 days Fekel et al., 1993
Tp 2446583.18300 \Sigma0.372 HJD Fekel et al., 1993
e 0.094 \Sigma0.011 Fekel et al., 1993
! 11.9 \Sigma6.5 Fekel et al., 1993
fl \Gamma21.869 \Sigma0.095 Fekel et al., 1993
K 12.56 \Sigma0.13 93.7 \Sigma2.5 km/s Fekel et al., 1993
Flux solution
F bol 20.5 \Sigma3.08 0.71 \Sigma0.071 \Theta10 \Gamma9 erg/cm 2 /s observed fluxes
F \Lambda 26.1 \Sigma3.92 2.11 \Sigma0.211 \Theta10 \Gamma9 erg/cm 2 /s dereddened fluxes
T eff 4500 \Sigma300 31500 \Sigma1500 K fluxes + models
` 106 \Sigma15 0.62 \Sigma0.060 \Theta10 \Gamma11 radians
EB\GammaV 0.13 \Sigma0.03 0.13 \Sigma0.03 fluxes + models
[Fe/H] \Gamma0.3 \Sigma0.3 \Gamma1.0 \Sigma0.5 spectrum + models
r s =rp 0.0058 \Sigma0.0010
Light curve solution
P 20.66096 \Sigma0.00104 days 1987/93 eclipses (1400 š A)
Tx 2449243.24590 \Sigma0.00180 HJD 1993 eclipse (1400 š A)
OE ec 0.06663 \Sigma0.00009 cycles 1993 eclipse (1400 š A)
OE in 0.00092 +0:00009
\Gamma0:00029 cycles 1993 eclipse (1400 š A)
Ü=r 0.0345
i 90.0 \Sigma5.3 degrees
r=a 0.2078 \Sigma0.0168 0.0012 \Sigma0.0002
Orbital dimensions
f(m) 0.00424 \Sigma0.00008 1.762 \Sigma0.081 M fi
a sin i 5.127 \Sigma0.053 38.249 \Sigma1.021 R fi
m sin 3 i 2.266 \Sigma0.167 0.304 \Sigma0.015 M fi
a 43.376 \Sigma1.241 R fi
mp=m s 7.460 \Sigma0.213
Stellar dimensions
m 2.27 \Sigma0.17 0.304 \Sigma0.015 M fi
r 9.01 \Sigma0.77 0.052 \Sigma0.010 R fi
L 30.3 \Sigma5.5 2.4 \Sigma0.7 L fi
log g 2.9 \Sigma0.1 6.5 \Sigma0.3 log cm/s 2
log gsp 2.5 7.2 \Sigma0.3 log cm/s 2 Randich et al. 1993, Lyff
d 192 \Sigma32 pc
Rotation
P rot 26 \Sigma1 days Hooten & Hall, 1990
v sin i 15 \Sigma2 km/s Fekel et al., 1993
r min 7.7 \Sigma1.1 R fi (solid body)
5.1 Analysis of the cool component
Indicators to the effective temperature of the cool star are
its spectral type and flux distribution. The IUE­LWP and
UBVRI photometry have been reexamined with the addition
of the 12¯m IRAS fluxes (Busso et al. 1988), 4.8¯m and
10.2¯m fluxes from the IRTF (Joseph 1991, see Appendix 1)
and JHKL photometry from SAAO (Whittet et al. 1995, see
Appendix 1). In­eclipse data from IUE­LWP, U and B were
used as by Jeffery et al. (1992); the eclipse is not observed
at longer wavelengths. Kurucz (1991a) model atmospheres
([Fe=H] = 0:0; \Gamma0:3; \Gamma1:0; log g = 2:5) were normalized to
match the dereddened cool­star fluxes and thus determine
the effective temperature. Angular radii were then obtained
by integrating the observed fluxes (supplemented by model
fluxes where necessary) to obtain an apparent luminosity,
which was then compared with the effective temperature.
Several points became clear from this exercise. First, the
flux distribution does not uniquely specify both the cool­star
effective temperature and reddening, but by adopting one,
the other is constrained. Second, if the reddening implied
by the hot­star analysis is adopted (EB\GammaV = 0:13, see next
section), the effective temperature of the cool star is too low
compared with its spectral type (K0III). Conversely, if the
effective temperature implied by the spectral type (4820 K,
Bell & Gustafsson 1989) is adopted, the required reddening
(EB\GammaV = 0:32) is much larger than that for the hot star.
Third, the UV fluxes strongly indicate that the K star is
metal­deficient ([Fe=H] ¸ \Gamma1:0), possibly in agreement with
the result for the subdwarf (see next section).
The prerequisite for the light­curve analysis is a reli­
able effective temperature for the K star and consistency
between the reddening towards both components. Adopting

6 C.S.Jeffery & T.Simon
EB\GammaV = 0:13 from the hot star analysis, T eff = 4500 \Sigma 100K
is obtained. The agreement between observed and theoreti­
cal fluxes improves with decreasing [Fe/H]. For [Fe=H] = 0:0,
the model predicts too little flux in both UV and IR when
normalized at R, but modifying T eff to improve one is at the
expense of the other. For [Fe=H] = \Gamma1:0, the agreement is
excellent (Fig. 2).
Such a low metallicity appears to be at odds with
the optical metal­line spectrum (Fekel 1996, priv. comm.),
although Randich et al. (1993) concluded that [Fe=H] =
\Gamma0:45. These authors also obtained T eff = 4500K; log g = 2:5
from a spectral­synthesis analysis. Since the metallicity of
chromospherically active stars is notoriously difficult to mea­
sure, a value of [Fe=H] = \Gamma0:3 is adopted as a compromise.
The same T eff is obtained from the flux distribution what­
ever the adopted [Fe/H].
Additional UBV(RI)C measurements were made nearly
simultaneously with the JHKL photometry (Roberts 1995,
see Appendix 1). The UBV(RI)C photometry indicates a
brighter star than does Fekel's (1986) photometry, but are
not fully consistent with the JHKL data. By carefully select­
ing which photometry to use, an alternative solution with
T eff = 4800K; log g = 2:5; [Fe=H] = 0:0 can be obtained.
This result is more consistent with the spectral type but
seems contrived. Clearly a study of the complete flux distri­
bution as a function of rotational phase would yield impor­
tant results.
Whichever values for [Fe/H] and T eff are finally shown
to be appropriate, the maxiumum uncertainties in total
observed flux and effective temperature are ¸ 15% and
¸ 7% respectively. These maximum error estimates have
been propagated in the orbital analysis given below.
5.2 Analysis of the hot component
The data available for measuring the hot­star parameters
are IUE­SWP and IUE­LWP photometry, two noisy IUE­
SWP­HIRES spectra and an IUE­SWP­LORES spectrum
obtained through the small aperture (SAP). The IUE­LWP
photometry of the B star is obtained by subtracting in­
eclipse photometry of the K star from the out­of­eclipse
photometry of the two stars combined. The two IUE­SWP­
HIRES spectra obtained by Jeffery et al. (1992) were ve­
locity shifted, bringing the stellar lines to the rest­frame,
coadded, smoothed (Gaussian filter: 0.25 š A FWHM) and nor­
malized to the continuum. Interstellar lines are diluted by
the coaddition and smoothing procedure. Only the wave­
length interval 1240 ! –= š A ! 1360 has sufficiently high
S/N to be useful, but contains T eff ­sensitive lines of C ii,iii
and Si ii,iii. The LORES­SAP spectrum provides a reason­
able Ly­ff absorption profile relatively uncontaminated by
geocoronal emission except in the line core. All IUE­SWP­
LORES images have been re­extracted from the IUE archive
using NEWSIPS software.
Since the Kurucz (1991a) model atmospheres only ex­
tend to log g Ÿ 5:0, a new grid of LTE line­blanketed model
atmospheres was calculated with the program STERNE
(Jeffery & Heber 1992) for [Fe=H] = 0:0; \Gamma1:0, log g =
5:0; : : : ; 8:0 and T eff = 25; : : : ; 35kK. These were com­
pared with the IUE flux distributions by normalizing
the models at – = 1500 š A and minimizing the quantity
jhFIUE(–)=Fmod(–)i \Gamma 1j (Fig. 3). With T eff ? 25 000K, the
Figure 3. The out­of­eclipse flux distribution of HD185510 B
(IUE SWP and LWP) dereddened with EB\GammaV = 0:13 and the
flux distribution for a model atmosphere with T eff = 31500 K,
log g = 7:2, [Fe=H] = \Gamma1:0.
IUE photometry does not yield a unique T eff but provides
a relationship between T eff , log g, and EB\GammaV , which can be
calibrated empirically. In order to obtain an independent
measure of EB\GammaV for the B star, additional diagnostics for
T eff and log g were sought.
The LTE spectral­line synthesis program SPECTRUM
(Dufton, Lennon, Conlon & Jeffery, unpublished) was ex­
tended to allow for the calculation of hydrogen Lyman lines
using tables (Lemke 1995, private comm.) based on the
broadening theory of Vidal et al. (1973). Line lists suitable
for synthesizing limited regions of UV spectrum based on the
extensive compilation of Kurucz (1991b) were also prepared
for use with SPECTRUM. Critical f­values were verified us­
ing the Opacity Project database (Cunto & Mendoza 1992).
Since the IUE­SWP­HIRES spectrum is still noisy, and
all lines are heavily blended, it was not appropriate to carry
out a line­by­line fine analysis of the spectrum. Rather the
best qualitative results were obtained from a systematic se­
quence of comparisons between synthetic and observed spec­
tra.
Synthetic spectra including all carbon and silicon lines
in the wavelength interval 1240 ! –= š A ! 1360 were calcu­
lated for each model atmosphere in the grid. These were con­
volved with a Gaussian (oe = 0:1 š A) to match the smoothing
carried out on the noisy IUE­HIRES spectra. Both atomic
species have strong lines from two adjacent ionization stages
in this wavelength region. It was immediately apparent, es­
pecially from C iii 1247 š A and Si iii 1295,. . . ,1304 š A that the
solar abundances for these species were too high; reducing
both by 1.0 dex led to a more satisfactory result. For each
value of log g, a value of T eff was determined which gave the
best comparison between lines from both ions. The quality
of the fit between observation and a final synthetic spec­
trum, which includes all of the strongest lines of C, N, O,
Mg, Al, Si, P, S, Ti, Mn, Cr and Fe, is only apparent in the
strong C and Si lines which dominate this region (Fig. 4).
Given the signal­to­noise ratio of the unsmoothed spectra,
it is difficult to be satisfied that any of the other features are
genuine absorption lines. The analysis of good­S/N GHRS
and IUE spectra of the hot subdwarf BD+75 ffi 325 by Hubeny

Timing the eclipse of HD185510 7
Figure 4. Synthetic spectra for HD185510 B in the interval 1240 ­ 1360 š A, compared with the combined IUE­SWP­HIRES spectrum.
All spectra have been degraded to a resolution of 0.25 š A(FWHM). The upper synthetic spectrum was computed with [Fe/H]= \Gamma1:0,
the lower with [Fe/H]= 0:0, but [C/H]=[Si/H]= \Gamma1:0. The synthetic spectra are duplicated in the upper part of each panel with a
flux­magnification of four.

8 C.S.Jeffery & T.Simon
Figure 5. A synthetic Lyman ff profile calculated for a model
atmosphere with T eff = 31500 K, log g = 7:2, [Fe=H] = \Gamma1:0 and
convolved with an interstellar hydrogen profile calculated assum­
ing EB\GammaV = 0:13 is compared with the IUE­LORES­SAP spec­
trum of HD185510 B. The line core is slightly contaminated by
geocoronal Ly­ff emission. The dashed line shows the Ly­ff profile
for T eff = 30000 K, log g = 6:0, [Fe=H] = \Gamma1:0 and EB\GammaV = 0:10
et al. (1991), which uses the same line list adopted here, also
misses several GHRS and IUE lines, suggesting that the line
list is far from complete and that there are many unresolved
and unidentified lines in our spectrum of HD185510. Hence
the overall [Fe/H] value is not determined. It need not reflect
the [C/H] and [Si/H] abundances since, in hot high­gravity
stars, gravitational settling and selective radiative forces can
seriously modify the atmospheric composition.
Theoretical hydrogen Ly­ff profiles were calculated for
the same grid of model atmospheres for comparison with the
LORES­SAP spectrum. Jeffery et al. (1992) obtained a good
correspondence at T eff = 25 000 K, log g = 6:0 using theo­
retical profiles based on pure hydrogen model atmospheres
(Wesemael et al. 1980). Their comparison could equally have
yielded T eff = 30 000 K, log g = 7:0, requiring EB\GammaV ¸ 0:1.
The new model profiles are weaker owing to the presence
of other background opacity in the atmosphere, implying
higher log g, but must be corrected for interstellar neutral
hydrogen absorption, for which the prescription given by
Groenewegen & Lamers (1989) was adopted. The hydro­
gen column N(H) can be obtained from a suitable N(H) --
EB\GammaV relation (Bohlin et al. 1978) since, given T eff and log g,
EB\GammaV is defined by the flux distribution. The position of the
local continuum, the Ly­ff line wings and other stellar ab­
sorption features in the SAP spectrum were verified from
the high­S/N coadded LAP spectrum. The SAP spectrum
served to define the profile nearer to the line core, but geo­
coronal Ly­ff contamination persists in the line centre. The
model profiles were required to match the adopted contin­
uum and the upper envelope defined by the SAP spectrum
within regions ! \Sigma25 š A and ? \Sigma5 š A of the line centre. The
value of log g finally obtained was higher than anticipated
and further gravity diagnostics from high­S/N UV spectra
remain a priority (Fig. 5).
Together, the three diagnostics of flux distribution,
C ii/iii and Si ii/iii ionization and the Ly­ff profile provide,
in principle, sufficient information to determine EB\GammaV , T eff
Figure 6. The loci of Ly­ff, EB\GammaV and ionization equilibrium so­
lutions in the T eff --log g plane, showing the final solution adopted
for HD185510 B.
Figure 7. The normalized flux (F 1400+2600 ) from HD185510
(secondary) during 1993 eclipse and the synthetic light curve com­
puted for the adopted orbital solution. The inset shows ingress
and egress at higher resolution assuming a purely geometric
eclipse (i = 70 ffi , dotted curve) and an atmospheric contribution
(linear, Ü=rp = 0:0345, i = 90 ffi solid curve).
and log g uniquely (Fig. 6). From the above, these are T eff =
31500 \Sigma 1500K, log g = 7:2 \Sigma 0:3 and EB\GammaV = 0:13 \Sigma 0:03.
The errors in T eff and log g were derived from the ionization
equilibrium and Ly­ff fits respectively.
6 ORBITAL ELEMENTS AND STELLAR
DIMENSIONS
In previous work, many fundamental data for HD185510
have already been derived, the remaining uncertainty being
that of the orbital inclination i.
The basic data required to construct a synthetic light
curve are orbital period P , mass ratio q, primary and sec­
ondary radii (rp =a and rs=a, where a is the orbital separa­
tion), and i. P and q are already established (Fekel et al.
1993); rp is defined, as a function of i, by OE ec . The angular
diameters of both stars and thus rs=rp have been obtained

Timing the eclipse of HD185510 9
above (`p ; `s ). In a system such as HD185510, tidal and rota­
tional distortion of both components is negligible (Hilditch
1995, priv. comm.) and the orbital eccentricity is small, thus
i and rp can be obtained analytically from OE in .
With the additional assumptions that the stars are uni­
formly illuminated with a discrete edge, the orbital inclina­
tion was established to be i = 70:3 ffi \Sigma 2:8 ffi . Hence stellar
dimensions of high accuracy could be obtained, but these
led to significant contradictions. First the radius for the cool
star derived from the light­curve solution was 18:0 \Sigma 1:8R fi ,
whereas the radius indicated by the apparent photometric
rotation period and rotational broadening in the spectral
lines is 7:6 \Sigma 1:1R fi = sin i. Second, the surface gravity in­
dicated for the hot subdwarf was 1.2 dex lower than that
indicated spectroscopically. Considerable re­examination of
all data and methods could yield no obvious solution except
that the discrete­disk approximation might be inappropri­
ate.
Limb­darkening on the subdwarf will modify the light­
curve to a small extent and in the wrong sense, effectively
by undersampling the disk and leading to an overestimate
of the inclination if ignored. For example, if 99% of the light
should arise within 80% of the disk radius, the above value
for i is reduced by ¸ 2:5 ffi .
More seriously, the eclipse ingress duration increases if
the geometrical extent of the cool­star atmosphere is a sig­
nificant fraction of the cool­star radius. Weak evidence for a
wavelength dependence of eclipse ingress and egress timings
has been noted already. In studies of the K giants HR6902A
(Schr¨oder et al. 1996) and Arcturus (Short 1995), density
scale heights significantly greater than 1% of the stellar ra­
dius have been determined. The geometrical density scale
height (Ü ), which we have adopted here to be similar to
the optical scale height (1400 š A), in an ATLAS9 model at­
mosphere with T eff = 4250, log g = 2:0, [Fe=H] = \Gamma0:5 is
Ü ¸ 0:1 Gm, or ¸ 0:018rp if rp = 8R fi . The eclipse due to
a linear atmosphere of such dimension increases the derived
inclination to ¸ 81 ffi .
An alternative means to obtain i is to assume the ro­
tation radius for the cool star and hence derive an effec­
tive scale height for the atmosphere. However with i = 90 ffi ,
which requires Ü=rp ¸ 0:03, the K­giant eclipse radius re­
mains greater than the rotation radius by 1:4R fi , although
the difference is now less than the combined observational
errors . The discrepancy is reduced by adopting a slightly
longer rotation period (26 d: Hooten & Hall 1990, 26.4 d:
Fekel et al. 1993) and almost entirely removed by increasing
the measured v sin i by only 1­oe to 17 km/s. This would be
consistent with an overestimate of the macroturbulent veloc­
ity (3 km/s, Fekel et al. 1993). By contrast, the minimum
eclipse radius is much more tightly constrained.
A further indication for a large inclination is given by
the subdwarf gravity; with i = 90 ffi , log g = 6:5 rather than
6.0 (i = 70 ffi ) and is considerably closer to the spectroscopic
value of 7.2. The residual discrepancy could be further re­
duced if the continuum were shown to have been overesti­
mated, although steps were taken to ensure its correct place­
ment (Fig. 5).
It will be seen that without substantially better eclipse
data, it is not possible to solve directly for i and/or Ü . How­
ever there are strong indications that i ? 80 ffi and most likely
i ¸ 90 ffi . Assuming the latter, the resulting solution is given
in Table 2, whilst the corresponding light curve, including
the eclipse due to a linear atmosphere of height Ü=rp = 0:034
is shown in Figure 7, superimposed on the IUE photometric
observations. z
Errors in Table 2 were propagated from the radial­
velocity data, from the measured flux distribution and the
adopted effective temperatures. A nominal error of \Sigma10%
has been assigned to the total flux from the hot star,
which was estimated by supplementing the measured IUE
fluxes with fluxes from the best­fit model atmosphere. The
\Sigma15% error in the cool­star flux has been discussed above.
The error in the hot star T eff is based on the size of the
T eff \Gamma log g \Gamma EB\GammaV domain where a reasonable fit of the
model flux distributions and theoretical Ly­ff profiles is ob­
tained (Fig. 6). The error of \Sigma5 ffi in i arises by assuming an
atmospheric scale height sufficient to derive i = 90 ffi
The orbital solution provides for the subdwarf a radius
which is less secure than that for the cool star. Whilst the
primary radius is tightly constrained by the eclipse duration,
the secondary radius relies on the measurement of relative
radii and OE in . Notice that the geometrical height of the cool­
star atmosphere is greater than the diameter of the subdwarf
and will dominate the eclipse geometry. Higher­precision UV
photometry with improved time and spectral resolution is
necessary to address these issues. Nevertheless the subdwarf
radius is sufficiently secure that the inconsistency with the
spectroscopic gravity must be resolved. A spectroscopic so­
lution with T eff = 31 000 \Sigma 1 500K, log g = 6:5 \Sigma 0:3 and
EB\GammaV = 0:10 \Sigma 0:03 would be consistent with all indica­
tors except Ly­ff and has a negligible effect on the orbital
analysis. Further insight into the spectroscopic gravity of
HD185510B would be rewarding.
7 NATURE OF THE HOT STAR IN HD185510
The aim of this investigation has been to determine whether
HD185510B is or will become a helium white dwarf, or
whether it is more properly analagous to the sdB, sdOB
and extreme horizontal­branch stars, being helium main­
sequence stars with T eff in the range 25 000 to 40 000 K.
The spectroscopic (Ly­ff) gravity for the hot subdwarf
(HD185510B) is sufficiently high that it is securely on the
cooling track for a 0:3M fi helium white dwarf (cf. Iben &
Tutukov 1986). The cooling time for such a white dwarf to
fade to a luminosity of ¸ 1L fi is ¸ 1 \Theta 10 7 yr, and the total
time from leaving the main sequence, which may include one
or more hydrogen­shell flashes, will not exceed ¸ 1 \Theta 10 8 yr.
This contradicts the expected ¸ 2 \Theta 10 8 yr which it would
take the K star to evolve to its present position from the
main sequence. Thus both the argument from evolutionary
ages and from the eclipse analysis argue against the high
surface gravity, so HD185510B is probably not currently a
helium white dwarf.
The mass of HD185510B is such that it may ignite 3ff
burning reactions and become a helium main­sequence star,
z The shorter ingress observed at 1800 š A gives a discrete­disk so­
lution with i = 78 ffi but which remains inconsistent with the rota­
tion radius. An atmospheric­eclipse solution with i = 90 ffi requires
an optical­depth scale height at this wavelength of only 0:012rp .

10 C.S.Jeffery & T.Simon
Figure 8. The location of HD185510 (secondary) in the log L \Gamma
log T eff (upper) and log g \Gamma log T eff (lower) diagrams. Both spec­
troscopic (Ly­ff) and binary orbit (r s =rp ) solutions for the radius
are indicated, with T eff adjusted to maintain consistency with
the spectroscopic solution. Other features of the upper diagram
are the same as in Fig. 5 of Jeffery et al. (1992), and the details
have been transposed to the log g \Gamma log T eff diagram, together with
observations of other sdB stars (open circles: Heber et al. 1984,
Heber 1986) and extreme horizontal­branch stars in NGC6752
(filled circles: Heber et al. 1986).
although with i = 90 ffi the situation is marginal. Following
He­ignition it would evolve to become a CO white dwarf
rather than a helium white dwarf. The binary eclipse radius
places HD185510B slightly below the helium main sequence
and, if the inclination is sufficiently low, on the lower fringe
of the region (in log g \Gamma T eff space) occupied by the sublu­
minous B and OB (sdB, sdOB) stars (Heber 1986, Moehler
et al. 1990). It is too overluminous for its mass to be a pure
helium main­sequence star, but the presence of a hydrogen­
burning shell could provide the necessary additional lumi­
nosity, the question being whether it could also provide the
necessary high surface temperature. Therefore HD185510B
may possibly be an sdB star, particularly if i ! 90 ffi .
The canonical mass for single sdB stars and EHB stars
of ¸ 0:45 \Gamma 0:55M fi is based on the coincidence of the ex­
tended horizontal branch (EHB) with the helium main se­
quence at ¸ 0:5M fi and with the measurement of the lumi­
nosities and gravities of EHB stars in NGC 6752 (Heber
et al. 1986). In the latter case, the individual EHB star
masses should be examined carefully. They range from 0:21\Gamma
1:26M fi , having a strongly asymmetric distribution. Three
EHB stars below the Newell gap (star nos. 3­118, 3675 and
3781) have a mean mass of 0:82 \Sigma 0:40M fi , whereas four blue
horizontal­branch (BHB) stars above the Newell gap (2167,
1083, 1056, 1007) have a mean mass of 0:39 \Sigma 0:14M fi . A
similar situation pertains to the BHB stars in M15 (Moehler
et al. 1992). These masses are very sensitive to the original
photometry and the adopted bolometric correction. Whilst
the reasonable assumption that errors in individual masses
average out provides a result consistent with the assumed
mass of 0:5M fi , there remains the possibility that the true
spread in sdB/sdOB masses may include the value observed
for HD185510B. The question that then arises is not whether
HD185510B can be identified with the sdOB stars in the
field but, since binary mass transfer can easily account for
the present instance, whether single stars can lose so much
mass as to resemble HD185510B.
At least one sdOB star, SB 707, has surface properties
(T eff = 34 000; log g = 6:0; [C=H] ! 0, Heber et al. 1984)
similar to those given by the low­inclination binary­eclipse
solution for HD185510B. Indeed, C and Si deficiencies seem
to be common amongst hot subdwarfs, whilst other species
(N) are near normal (Lamontagne et al. 1985). Thus there
are good empirical grounds for supposing HD185510B to
be indistinguishable from the sdOB stars. The philosophical
problem is to demonstrate how a star of such low mass can
have the observed effective temperature and luminosity.
However the status of HD185510B as an sdB/sdOB star
is not proven by the foregoing arguments, and the earlier
conclusion that HD185510B could be a nascent helium white
dwarf (Jeffery et al. 1992) remains a significant possibility.
8 CONCLUSION
A greatly improved ultraviolet light curve for HD185510 has
allowed many system parameters to be obtained with con­
siderable precision. The system consists of a K giant primary
with T eff = 4 500 \Sigma 300K and a hot subdwarf or white dwarf
secondary with T eff = 31 500 \Sigma 1 500K, with relative radii
r s =rp = 0:0058 \Sigma 0:0010. Eclipse measurements have deter­
mined the times of contact to \Sigma0:0001 cycles, or ¸ \Sigma3m.
There is evidence that the subdwarf is partially eclipsed by
a more extended atmosphere around the K giant, such as in
a i Aurigae system. This, and a contradiction between the
radii of the K giant inferred from the eclipse geometry and
from the rotational period and v sin i, continue to prevent
an unambiguous measurement of the inclination i. Whilst
the body of evidence favours a high inclination i ¸ 90 \Sigma 5 ffi ,
yielding masses of 0:304 \Sigma 0:015M fi and 2:27 \Sigma 0:17M fi for
the hot and cool components respectively, an inclination as
low as i ¸ 80 ffi cannot be ruled out.
The available data suggest that HD185510B is probably
not currently a helium white dwarf, but resembles better
an extreme sdB/sdOB star. There remains some possibility
that it is a nascent helium white dwarf.
This investigation has exhaustively examined the avail­
able photometry and spectroscopy of HD185510 in an effort
to resolve the nature of its components. It is ultimately the
limitations of these data and the nature of the stars them­
selves which have prevented a more successful and definitive

Timing the eclipse of HD185510 11
outcome. More detailed work is required in the following
areas.
(i) High­resolution high­S/N UV spectroscopy of the sub­
dwarf will allow its composition, T eff and log g to be mea­
sured more precisely. In particular the metal abundances
suggest differential radiative levitation is operating and re­
quire close examination.
(ii) Still higher time­resolution UV spectrophotometry of
eclipse ingress and egress are needed to determine the geo­
metric and optical structure of the K­star photosphere.
(iii) A detailed study of the K­star photosphere is also im­
portant in order to determine the true mean photospheric
T eff and angular diameter. This is complicated by the contri­
bution of spots which demands that multi­wavelength pho­
tometry be carried out around the 25­day rotation period.
(iv) The discrepancy between the spectral­type T eff
(4 800K) and bolometric flux T eff (4 500K) of the K star
has been attributed to metallicity. This needs to be verified
from high­resolution optical and infrared spectroscopy.
Given the intriguing nature of HD185510 as a post­Algol
binary, the empirical measurements of mass­loss and mass­
exchange that can be derived from such a well­defined sys­
tem, and the nature of the individual stars themselves, these
investigations are well­worth pursuing.
Acknowledgments
The authors are indebted to observatory staff at VILSPA
and Goddard for help in making these observations, and in
particular to fellow IUE observers Dave Stickland and Carol
Ambruster for enabling us to accommodate last­minute ad­
justments to the orbital ephemeris and to Bob Joseph, Doug
Whittet, Dave Kilkenny and Greg Roberts for obtaining the
infrared photometry. Ron Hilditch, Ian Short, Bill Napier
and Brendan Byrne took part in countless but vital conver­
sations addressing the cool­star radius dilemma.
REFERENCES
Balona L.A., Lloyd Evans T., Simon, T., Sonneborn G., 1987.
IBVS No. 3061
Bell R.A., Gustafsson B., 1989. MNRAS 236, 653
Bohlin R.C., Savage B.D., Drake J.F., 1978. ApJ 224, 131
Busso M., Scaltriti F., Persi P., Farrari­Toniolo M., Origlia L.,
1988. MNRAS 234, 445
Cunto W., Mendoza C., 1992. Rev. Mexicana Astron.Astrofis. 23,
107
Fekel F.C., Moffett T.J., Henry G.W., 1986. ApJS 60, 551
Fekel F.C., Henry G.W., Busby M.R., Eitter J.J., 1993. AJ 106,
2370
Giddings J.R., Rees P.C.T.R., Mills D., Clayton M., 1994. Star­
link User Note 37
Groenewegen M.A.T., Lamers H.J.G.L.M., 1989. AAS 79, 359
Heber U., 1986. AA 155, 33
Heber U., Hunger K., Jonas G., Kudritzki R.P., 1984. AA 130,
119
Heber U., Kudritzki R.P., Caloi V., Castellani V., Danziger J.,
Gilmozzi R., 1986. AA 162, 171
Hooten J.T., Hall D.S., 1990. ApJS 74, 225
Hubeny I., Heap S.R., Altner B., 1991. ApJ 377, L33
Iben I., Jr., Yutukov A.V., 1986, ApJ 311, 742
Jeffery C.S., Simon T., Lloyd Evans T., 1992. MNRAS 258, 64
Jeffery C.S., Heber U., 1992. AA 260, 133
Johnson H.L., 1966. Ann.Rev.Astr.Astrophys. 4, 193
Kurucz R.L., 1991a. in 'Precision Photometry: Astrophysics of
the Galaxy', eds. A.G.Davis Philip, A.R.Upgren, K.A.Janes,
L.Davis Press, Schenectady
Kurucz R.L., 1991b. in 'Stellar Atmospheres: Beyond Classi­
cal Models', p.441, eds. L.Crivellari, I.Hubeny, D.G.Hummer,
NATO ASI Series Vol. C341, Kluwer, Dordrecht
Lamontagne R., Wesemael F., Fontaine G., Sion E.M., 1985. ApJ
299, 496
Moehler S., de Boer K.S., Heber U., 1992. 'The Atmospheres of
Early­Type Stars', p.251. eds. Heber U., Jeffery C.S., Springer
Verlag.
Press W.H., Flannery W.P., Teukolsky S.A., Vetterling W.T.,
1989. Numerical Recipes. Cambridge University Press
Randich S., Gratton R., Pallavicini R., 1993, AA 273, 194
Schr¨oder K.­P., Marshall K.P., Griffin R.E.M., 1996. AA 311, 631
Short C.I., 1995. Ph.D.Thesis, University of Toronto
Vidal C.R., Cooper J., Smith E.W., 1973. ApJS 25, 37
APPENDIX A: VISUAL AND INFRARED
PHOTOMETRY
The photometric indices and fluxes used in the analysis of
the K­star effective temperature are shown in Table 1. M
and N indices were kindly obtained at the IRTF by Dr
R.D.Joseph on the night of 12/13 September 1991, rela­
tive to ff Her for which the standard values (IRTF Hand­
book) of M= \Gamma3:45 and N= \Gamma3:92 have been adopted.
JHKL photometry was obtained at SAAO in 1995 April
(JD 2449823.65 and 2449825.67) by Drs D.Whittet and
D.Kilkenny.
Dr G.Roberts obtained UBV(RI)C photometry, also
from SAAO in 1995 April (JD 2449820.63 and 2449825.65).
The (RI)C measures should not be directly compared with
the Johnson RI measures. Roberts' UBV measurements
showed HD185510 to be brighter than observed by Fekel
et al. (1986). The differences between Roberts' and Fekel's
measurements are within the 0:2 mag full­amplitude V vari­
ations noted by Hooten & Hall (1990) and can probably be
attributed to the surface spot distribution at the times of
observation. A more detailed discussion of the K­star fluxes
will be pursued elsewhere. Balona et al.'s (1987) in­eclipse
data (U and B) were used in the current analysis.
Magnitudes were converted to fluxes (erg/cm 2 /s/ š A) us­
ing
f– = 10 0:4(C – \Gammam – ) (1)
where the adopted scale factors C– (from Johnson 1966) are
also given in Table 1.
This paper has been produced using the Royal Astronomical
Society/Blackwell Science L A T E X style file.

12 C.S.Jeffery & T.Simon
Table 1. Visual and infrared photometric indices for HD185510
–c C – m –
normal eclipse
Joh66 Fek86 Rob95 Bal87
U 3600 š A ­20.90 10.33 10.17 10.50
B 4400 š A ­20.36 9.63 9.48 9.65
V 5500 š A ­21.02 8.50 8.33
R 7000 š A ­21.89 7.51 (7.68)
I 9000 š A ­22.70 6.84 (7.05)
Whi95
J 1.25¯m ­23.67 6.16
H 1.65¯m ­24.80 5.49
K 2.2¯m ­26.02 5.36
L 3.4¯m ­27.73 5.25
Jos91
M 4.8¯m ­29.14 5.19
N 10.2¯m ­32.28 5.28
Bus88 Bus88
[12] 12.0¯m ­33.07 4.94
Notes: Bal87: Balona et al. 1987, Bus88: Busso et al. 1988,
Fek86: Fekel et al. 1986, Joh66: Johnson 1966, Jos91: Joseph 1991
(priv. comm.), Rob95: Roberts & Kilkenny 1995 (priv. comm.,
UBV(RI)C ), Whi95: Whittet & Kilkenny 1995 (priv. comm.).