1: \documentclass[12pt,preprint]{aastex}
2: \shorttitle{Cepheid Pulsations}
3: \shortauthors{Lane et al.}
4:
5: \begin{document}
6:
7: \title{Long Baseline Interferometric Observations of Cepheids.}
8:
9: \author{Benjamin\,F. Lane}
10: \affil{Department of Geological \& Planetary Sciences, MS
11: 150-21, California Institute of Technology, Pasadena CA 91125,
12: U.S.A.}
13: \email{ben@gps.caltech.edu}
14:
15: \author{Michelle J. Creech-Eakman}
16: \affil{Caltech/JPL Postdoctoral Scholar, Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, U.S.A. }
17: \email{mce@huey.jpl.nasa.gov}
18: \and
19: \author{Tyler E. Nordgren}
20: \affil{University of Redlands, Department of Physics, 1200 E Colton Ave, Redlands, California, U.S.A.}
21: \email{Tyler\_Nordgren@redlands.edu}
22:
23:
24: \begin{abstract}
25:
26:
27: We present observations of the galactic cepheids $\eta$ Aql and
28: $\zeta$ Gem. Our observations are able to resolve the diameter
29: changes associated with pulsation. This allows us to determine the
30: distance to the Cepheids independent of photometric observations. We
31: determine a distance to $\eta$ Aql of $ 320\pm32$ pc, and a distance
32: to $\zeta$ Gem of $ 362\pm38$ pc. These observations
33: allow us to calibrate surface brightness relations for use in
34: extra-galactic distance determination. They also provide
35: a measurement of the mean diameter of these Cepheids, which is useful
36: in constructing structural models of this class of star.
37: \end{abstract}
38:
39: \keywords{Cepheids-- stars:fundamental parameters--stars:individual ($\eta$ Aquilae, $\zeta$ Geminorum)}
40:
41: \section{Introduction}
42:
43: The class of pulsating stars known as Cepheids is a cornerstone in
44: determining the distances to nearby galaxies. This is because Cepheids
45: exhibit a well-behaved period-luminosity relation which can be locally
46: calibrated \citep{jacoby92}. In addition, these stars are massive and
47: thus intrinsically very luminous, making it possible to observe
48: Cepheids located in very distant galaxies \citep{tan99,feast99}.
49: Because of the usefulness and fundamental importance of Cepheids, it
50: is important to calibrate their period-luminosity relation. This has
51: been done using a variety of methods, including parallax \citep{esa97,
52: fc97}, Baade-Wesselink methods \citep{wesselink46,bersier97} and
53: surface brightness \citep{laney95,fg97,ripepi97}. The
54: period-luminosity relations used currently have uncertainties on the
55: order of 0.09 mag \citep{feast99}, which in turn make up a significant
56: portion of the systematic uncertainty in estimates to the Large
57: Magellanic Cloud.
58:
59: Using long-baseline stellar interferometry it is possible to resolve
60: the diameter changes undergone by a nearby Cepheid during a
61: pulsational cycle. When such diameter measurements are combined with
62: radial velocity measurements of the stellar photosphere, it is
63: possible to determine the size of and distance to the Cepheid. Such a
64: direct measurement is independent of photometric observations and
65: their associated uncertainties.
66:
67: The Palomar Testbed Interferometer (PTI) is located on Palomar
68: Mountain near San Diego, CA \citep{colavita99}. It combines starlight
69: from two 40-cm apertures to measure the amplitude (a.k.a. visibility)
70: of the resulting interference fringes. There are two available
71: baselines, one 110-m baseline oriented roughly North-South (hereafter
72: N-S), and one 85-m baseline oriented roughly North-Southwest (called
73: N-W). In a previous paper \citep{lane00} we presented observations using PTI
74: of the Cepheid $\zeta$ Gem. Here we report on additional
75: interferometric observations of $\zeta$ Gem, as well as a second
76: Galactic Cepheid, $\eta$ Aql. These observations allow us to determine
77: the distances to these Cepheids with the aim of reducing the
78: uncertainty in currently used period-luminosity relations for
79: Cepheids.
80:
81:
82: \section{Observations}
83:
84: We observed the nearby galactic cepheids $\eta$ Aql and $\zeta$ Gem on
85: 22 nights between 2001 March 13 and 2001 July 26. The observing
86: procedure followed standard PTI practice
87: \citep{boden98,colavita99}. For the observations of $\eta$ Aql the N-W
88: baseline was used, while observations of $\zeta$ Gem used the N-S
89: baseline. Each nightly observation consisted of approximately ten
90: 130-second integrations during which the fringe visibility was
91: averaged. The measurements were done in the $1.52-1.74~\mu$m
92: (effective central wavelength $1.65~\mu$m) wavelength region, similar
93: to the astronomical $H$ band. Observations of calibration sources
94: were rapidly (within less than $\sim 10$ minutes) interleaved with the
95: Cepheid observations, and after each 130-second integration the
96: apertures were pointed to dark sky and a 30-second measurement of the
97: background light level was made.
98:
99: The calibrators were selected to be located no more than 16 degrees
100: from the primary target on the sky and to have similar $H$-band
101: magnitudes. In choosing calibration sources we avoided known binary or
102: highly variable stars. The calibrators used are listed in Table
103: \ref{tab:calibs}. In this paper we make use of previously published
104: observations of the Cepheid $\zeta$ Gem \citep{lane00}. However, in
105: order to improve on the previously published results we carried out
106: additional observations of this source on 2001 March 13--15. We also
107: observed additional unresolved calibrators in order to reduce the
108: level of systematic uncertainty. The original data have been jointly
109: re-reduced using the improved calibrator diameters and
110: uncertainties. However, note that the primary calibrator diameter has
111: not changed from the value used in \citet{lane00}.
112:
113: \section{Analysis \& Results}
114:
115: \subsection{Fringe Visibilities \& Limb Darkening}
116: PTI uses either a 10 or 20 ms sample rate. Each such sample provides
117: a measure of the instantaneous fringe visibility and phase. While the
118: phase value is converted to distance and fed back to the active delay
119: line to provide active fringe tracking, the measured fringe visibility
120: is averaged over the entire 130-second integration. The statistical
121: uncertainty in each measurement is estimated by breaking the 130 second
122: integration into five equal-time segments and measuring the standard
123: deviation about the mean value.
124:
125: The theoretical relation between source brightness distribution and
126: fringe visibility is given by the van Cittert-Zernike theorem. For a
127: uniform intensity disk model the normalized fringe visibility (squared) can be
128: related to the apparent angular diameter as
129: \begin{equation}
130: V^2 = \left( \frac{2 \; J_{1}(\pi B \theta_{UD} / \lambda_0)}{\pi B\theta_{UD} / \lambda_0} \right)^2
131: \label{eq:V2_single}
132: \end{equation}
133: where $J_{1}$ is the first-order Bessel function, $B$ is the projected
134: aperture separation, $\theta_{UD}$ is the apparent angular diameter of
135: the star in the uniform-disk model, and $\lambda_0$ is the center-band
136: wavelength of the observation. It follows that the fringe visibility of a
137: point source measured by an ideal interferometer should be unity.
138: For a more realistic model that includes limb darkening one can derive a
139: conversion factor between a uniform-disk diameter ($\theta_{UD}$) and
140: a limb-darkened disk diameter ($\theta_{LD}$) given by \citep{welch94}
141: \begin{equation}
142: \theta_{UD} = \theta_{LD} \sqrt{1 - \frac{A}{3} - \frac{B}{6}}
143: \end{equation}
144: where A and B are quadratic limb darkening coefficients, determined by
145: the spectral type of the source \citep{claret95}. The limb darkening
146: correction factors ($k = \theta_{UD}/\theta_{LD}$) used for the
147: Cepheids are shown in Table \ref{tab:targ} and for the
148: calibrators in Table \ref{tab:calibs}.
149:
150: \subsection{Visibility Calibration}
151:
152: The first step in calibrating visibilities measured by PTI is to
153: correct for the effects of detector background and read-noise, the
154: details of which are discussed in \citet{colavita99} and
155: \citet{colavita99b}. However, the visibilities thus produced are not
156: yet final: due to a variety of effects, including systematic
157: instrumental effects, intensity mismatches, and atmospheric
158: turbulence, the fringe visibility of a source measured by PTI is lower
159: than that predicted by Eq. 1. In practice the system response function
160: (called the system visibility) is typically $\sim 0.75$ and
161: furthermore is variable on 30 minute timescales. Hence the
162: visibilities must be calibrated by observing sources of known
163: diameter.
164:
165: Determining the diameter of the calibration sources was a multi-step
166: process in which we made use of both models and prior observations.
167: For each Cepheid we designated a single, bright K giant as a primary
168: calibrator, which was always observed in close conjunction with the
169: target Cepheid (HD 189695 for $\eta$ Aql, and HD 49968 for $\zeta$
170: Gem). We used model diameter estimates for the primary calibrators
171: from previously published results based on spectro-photometry and
172: modeling \citep{cohen99}.
173:
174: In order to verify that the primary calibrators were stable and had
175: angular diameters consistent with the \citet{cohen99} results, we
176: observed them together with a number of secondary calibrators. These
177: secondary calibrators were typically less resolved than the primary
178: calibrators and hence less sensitive to uncertainties in their
179: expected angular diameter. However, they were fainter than the primary
180: calibrators, and tended to be located further away on the sky. For the
181: secondary calibrators an apparent diameter was estimated using three
182: methods: (1) we used available archival photometry to fit a black-body
183: model by adjusting the apparent angular diameter, bolometric flux and
184: effective temperature of the star in question so as to fit the
185: photometry. (2) We repeated the above fit while constraining the
186: effective temperature to the value expected based on the published
187: spectral type. (3) We estimated the angular diameter of the star
188: based on expected physical size (derived from spectral type) and
189: distance (determined by Hipparcos). We adopted the weighted (by the
190: uncertainty in each determination) mean of the results from the above
191: methods as the final model diameter for the secondary calibrators, and
192: the uncertainty in the model diameter was taken to be the deviation
193: about the mean.
194:
195: In addition to the model-based diameter estimates derived above we
196: also used extensive interferometric visibility measurements for the
197: primary and secondary calibrators; given that several of the
198: calibrators were observed within a short enough period of time that
199: the system visibility could be treated as constant, it was possible to
200: find a set of assumed calibrator diameters that are maximally
201: self-consistent, by comparing observed diameter ratios for which the
202: system visibility drops out. To illustrate, let $\theta_i$ be an
203: adjustable parameter, representing the diameter of star $i$. Let
204: $\hat{\theta}_i$ and $\sigma_{\hat{\theta_i}}$ be the theoretical
205: model diameter and uncertainty for star $i$ derived above, and let
206: $\tilde{R_{ij}}$ and $\sigma_{\tilde{R_{ij}}}$ be the
207: interferometrically observed diameter ratio and uncertainty of stars
208: $i$ and $j$. For notational simplicity, define $R_{ij}$ as the ratio
209: of $\theta_i$ and $\theta_j$. Define the quantity
210: \begin{equation}
211: \chi^2 = \sum_{i} \left[ \frac{ \hat{\theta}_i - \theta_i }{\sigma_{\hat{\theta_i}}} \right]^2 \
212: + \sum_{i} \sum_{j < i} \left[ \frac{ \tilde{R_{ij}} - R_{ij}}{\sigma_{\tilde{R_{ij}}}} \right]^2
213: \end{equation}
214: By adjusting the set of $\theta_i$ to minimize $\chi^2$ we produce a
215: set of consistent calibrator diameters, taking into account both input
216: model knowledge and observations. The resulting diameter values are
217: listed in Table \ref{tab:calibs}. Uncertainties were estimated using
218: the procedure outlined in \citet{press86} assuming normally
219: distributed errors.
220:
221: We verified that the primary calibrators were stable as follows: using
222: the secondary calibrators to calibrate all observations of the primary
223: calibrators we fit a constant-diameter, single-star, uniform-disk
224: model to the primary calibrators. In all cases the scatter about the
225: single-star model was similar to expected system performance
226: \citep{boden98}: for HD 189695, 21 points were fit, the average
227: deviation in $V^2$ was 0.035 and the goodness-of-fit parameter of the
228: line fit, $\chi^2$ per degree of freedom ($\chi^2_{dof}$, not to be
229: confused with Eq. 3 above), in the line fit was 0.46. For HD 49968, 82
230: points were fit, the average deviation was 0.038 and $\chi^2_{dof}$ =
231: 0.76.
232:
233: \begin{table}
234: \begin{center}
235: \begin{tabular}{ccccccc}
236: & & & & & & \\
237: \tableline
238: \tableline
239: Star & Alternate & Period & Epoch & Limb Dark. \\
240: Name & Name & (d) & JD & Factor (k) \\
241: \tableline
242: $\eta$ Aql & HD 187929 & 7.176711 & 2443368.962 & $0.97 \pm 0.01$ \\
243: $\zeta$ Gem& HD 52973 & 10.150079& 2444932.736 & $0.96 \pm 0.01$\\
244: \tableline
245: \end{tabular}
246: \caption{\label{tab:targ} Relevant parameters of the Cepheids. The limb darkening
247: factor is defined as $k = \theta_{UD}/\theta_{LD}$.}
248:
249: \end{center}
250: \end{table}
251:
252:
253: \begin{table}
254: \begin{center}
255: \begin{tabular}{lcccccccc}
256: & & & & & & & & \\
257: \tableline
258: \tableline
259: Calibrator & Spectral & Diameter Used & Limb Dark. & Used to & Cal. & Angular Sep. \\
260: & Type & $\theta_{UD}$ (mas) & Factor (k) & calibrate & Type & (deg) \\
261: \tableline
262: HD 189695 & K5 III & $1.89 \pm 0.07$ & $0.943\pm0.007$ & $\eta$ Aql & Pri. Cal & 7.8 \\
263: HD 188310 & G9.5 IIIb & $1.57 \pm 0.08$ & $0.955\pm0.007$ & $\eta$ Aql & Sec. Cal & 8.2 \\
264: HD 181440 & B9 III & $0.44 \pm 0.05$ & $0.975\pm0.007$ & $\eta$ Aql & Sec. Cal & 7.5 \\
265: HD 49968 & K5 III & $1.78 \pm 0.02$ & $0.939\pm0.006$ & $\zeta$ Gem & Pri. Cal & 4.1 \\
266: HD 48450 & K4 III & $1.94 \pm 0.02 $ & $0.949\pm0.007$ & $\zeta$ Gem & Sec. Cal & 9.5 \\
267: HD 39587 & G0 V & $1.09 \pm 0.04$ & $0.963\pm0.006$ & $\zeta$ Gem & Sec. Cal & 16 \\
268: HD 52711 & G4 V & $0.55 \pm 0.04$ & $0.962\pm0.006$ & $\zeta$ Gem & Sec. Cal & 8.8 \\
269:
270: \tableline
271: \end{tabular}
272: \caption{\label{tab:calibs} Relevant parameters of the calibrators. The angular separation
273: listed is the angular distance from the calibrator to the Cepheid it is used to calibrate. }
274:
275: \end{center}
276: \end{table}
277:
278: While analyzing the data it was noticed that during observations with
279: the N--W baseline of relatively low declination sources, such as
280: $\eta$ Aql and its calibrators, the stability of the interferometer
281: system visibility was strongly dependent on the hour angle of the
282: source: for observations of $\eta$ Aql obtained at positive hour
283: angles the scatter in the system visibility increased by a factor of
284: 2--3, while the mean value trended down by 20\%/hr. There are two
285: potential explanations for this effect: (1) for these observations the
286: optical delay lines are close to their maximum range, which can
287: exacerbate internal system misalignments and lead to vignetting. (2)
288: When observing low declination sources past transit, the siderostat
289: orientation is such that surface damage near the edge of one of the
290: siderostat mirrors causes vignetting. Thus it was decided to discard
291: observations of $\eta$ Aql taken at positive hour angles,
292: corresponding to $\sim 20$\% of the available data. We note that
293: including the data does not significantly change the final results
294: ($\sim 0.3\sigma$), it
295: merely increases the scatter substantially (for the pulsation fit
296: discussed below the goodness-of-fit parameter $\chi^2_{dof}$ increased
297: from 1.06 to 4.5).
298:
299: \subsection{ Apparent Angular Diameter}
300:
301: Once the measured visibilities were calibrated we used all the
302: available calibrated data from a given night to determine the apparent
303: uniform-disk angular diameter of the target Cepheid on that particular night
304: by fitting to a model given by Eq. 1. Results are given in
305: Tables \ref{tab:diams_ea} and \ref{tab:diams_zg} and plotted in
306: Fig. 1. Uncertainties were estimated based on the scatter about the
307: best fit. It should be noted that although $\eta$ Aql is known to have
308: a companion \citep{bv85} it is sufficiently faint (average $\Delta m_H
309: = 5.75$ mag) that it will have a negligible effect ($\Delta V^2 \sim
310: 0.005$) on the fringe visibilities measured in the H band.
311:
312: \begin{table}
313: \begin{center}
314: \begin{tabular}{ccc}
315: & \\
316: \tableline
317: \tableline
318: Epoch & Angular Diameter & No. Scans \\
319: JD-2400000.5 & $\theta_{UD}$ (mas) & \\
320: \tableline
321: 52065.420 & $ 1.654 \pm 0.011$ & 9 \\
322: 52066.414 & $ 1.654 \pm 0.017$ & 9 \\
323: 52067.405 & $ 1.694 \pm 0.040$ & 8 \\
324: 52075.383 & $ 1.740 \pm 0.027$ & 12 \\
325: 52076.384 & $ 1.799 \pm 0.014$ & 9 \\
326: 52077.372 & $ 1.822 \pm 0.021$ & 13 \\
327: 52089.350 & $ 1.715 \pm 0.019$ & 11 \\
328: 52090.354 & $ 1.798 \pm 0.020$ & 9 \\
329: 52091.346 & $ 1.764 \pm 0.022$ & 7 \\
330: 52095.360 & $ 1.567 \pm 0.049$ & 1 \\
331: 52099.337 & $ 1.800 \pm 0.025$ & 2 \\
332: 52101.329 & $ 1.632 \pm 0.037$ & 5 \\
333: 52103.293 & $ 1.656 \pm 0.040$ & 7 \\
334: 52105.300 & $ 1.798 \pm 0.024$ & 6 \\
335: 52106.283 & $ 1.816 \pm 0.016$ & 19 \\
336: 52107.302 & $ 1.809 \pm 0.027$ & 11 \\
337: 52108.308 & $ 1.702 \pm 0.032$ & 7 \\
338: 52116.276 & $ 1.611 \pm 0.023$ & 7 \\
339:
340: \tableline
341: \end{tabular}
342: \caption{\label{tab:diams_ea} The measured uniform-disk diameters of $\eta$ Aql.
343: The uncertainties are the statistical uncertainty
344: from the scatter during a night, and do not include systematic uncertainty
345: in the calibrator diameters; this adds an additional uncertainty of $0.07$ mas in the aggregate mean diameter. }
346:
347: \end{center}
348: \end{table}
349:
350:
351: \begin{table}
352: \begin{center}
353: \begin{tabular}{ccc}
354: & \\
355: \tableline
356: \tableline
357: Epoch & Angular Diameter & No. Scans\\
358: JD-2400000.5 & $\theta_{UD}$ (mas) & \\
359: \tableline
360: 51605.226 & $ 1.676 \pm 0.015$ & 15 \\
361: 51606.241 & $ 1.675 \pm 0.047$ & 3 \\
362: 51614.192 & $ 1.797 \pm 0.060$ & 7 \\
363: 51615.180 & $ 1.737 \pm 0.031$ & 10 \\
364: 51617.167 & $ 1.587 \pm 0.028$ & 10 \\
365: 51618.143 & $ 1.534 \pm 0.008$ & 11 \\
366: 51619.168 & $ 1.549 \pm 0.018$ & 15 \\
367: 51620.169 & $ 1.585 \pm 0.028$ & 15 \\
368: 51622.198 & $ 1.673 \pm 0.046$ & 6 \\
369: 51643.161 & $ 1.663 \pm 0.012$ & 9 \\
370: 51981.182 & $ 1.685 \pm 0.014$ & 23 \\
371: 51982.164 & $ 1.636 \pm 0.020$ & 16 \\
372: 51983.201 & $ 1.589 \pm 0.021$ & 15 \\
373: 51894.387 & $ 1.619 \pm 0.019$ & 13 \\
374: 51895.369 & $ 1.629 \pm 0.014$ & 12 \\
375:
376: \tableline
377: \end{tabular}
378: \caption{\label{tab:diams_zg} The measured uniform-disk diameters of $\zeta$ Gem.
379: The uncertainties are the statistical uncertainty
380: from the scatter during a night, and do not include systematic uncertainty
381: in the calibrator diameters; this adds an additional uncertainty of $0.024$ mas
382: in the aggregate mean diameter.
383: }
384:
385: \end{center}
386: \end{table}
387:
388: It is clear from Fig. \ref{diams} that the measured angular diameters
389: are not constant with time. Fitting a constant-diameter model to the data
390: produces a rather poor fit (see Table \ref{tab:result}). However, we
391: list the resulting mean angular diameters in order to facilitate comparison
392: with previous interferometric results.
393:
394: \subsection{Distances \& Radii}
395:
396: Determining the distance and radius of a Cepheid via the
397: Baade-Wesselink method requires comparing the measured changes in
398: angular diameter to the expansion of the Cepheid photosphere measured
399: using radial velocity techniques. In order to determine the expansion
400: of the Cepheid photospheres we fit a fifth-order Fourier series to
401: previously published radial velocities. For $\eta$ Aql we used data
402: from \citet{bersier02} as well as data published by \citet{jw81,jw87},
403: while for $\zeta$ Gem we used data from \citet{bersier94}. Both sets
404: of data were from measurements made at optical wavelengths. The
405: measured radial velocities were converted to physical expansion rates
406: using a projection factor (p-factor), which depends on the detailed
407: atmospheric structure and limb darkening of the Cepheid as well as on the
408: details of the equipment and software used in the measurement
409: \citep{hind86,albrow94}. It is important to note that the p-factor is
410: not expected to stay constant during a pulsational cycle. The exact
411: phase dependence of the p-factor is beyond the scope of this
412: paper. However, for $\eta$ Aql and $\zeta$ Gem, the net effect of a
413: variable p-factor can be approximated by using a 6\% larger constant
414: p-factor \citep{sabbey95}. Thus for both Cepheids we use an effective
415: p-factor of $1.43 \pm 0.06$, constant for all pulsational phases.
416:
417: We convert the radial velocity Fourier series into a physical size
418: change by integrating and multiplying by limb-darkening and p-factors.
419: Although the limb-darkening does vary with changing ${\rm T}_{eff}$
420: during a pulsational cycle, the effect is small: for $\zeta$ Gem $k$
421: varies from 0.960 to 0.967, i.e.\ less than the quoted uncertainty.
422: The size change can in turn be converted into an angular size model
423: with three free parameters: the mean physical radius, the distance to
424: the star, and a phase shift. The latter is to account for
425: possible period changes, inaccuracies in period or epoch, or phase
426: lags due to level effects (where the optical and infrared photospheres
427: are at different atmospheric depths; see below). We adjust the model
428: phase, radius and distance to fit the observed angular diameters.
429: Results of the fits for $\eta$ Aql and $\zeta$ Gem are given in Table
430: \ref{tab:result}.
431:
432: There are several sources of uncertainty in the above fits: in
433: addition to the purely statistical uncertainty there are systematic
434: uncertainties of comparable magnitude. The three primary sources of
435: systematic uncertainty are: uncertainty in the calibrator diameters,
436: uncertainty in the p-factor, and uncertainty in the limb darkening
437: coefficients. The magnitude of each effect was estimated separately by
438: re-fitting the model while varying by $\pm 1\sigma$ each relevant
439: parameter separately. The total systematic uncertainty was calculated
440: as
441: \begin{equation}
442: \sigma_{sys}^2 = \sigma_{cal}^2 + \sigma_{p-fac}^2 + \sigma_{limb dark.}^2
443: \end{equation}
444:
445: In order to explore the possibility of wavelength-dependent effects on
446: the measured radial velocity, e.g.\ due to velocity gradients in the
447: Cepheid atmospheres (``level effects''), we re-fit for the radius and
448: distance of $\eta$ Aql using a radial velocity curve based on radial
449: velocity data obtained at wavelengths of $1.1$ and $1.6 \mu$m by
450: \citet{sasselov90}. Because of the limited number of observations
451: available (e.g only 3 $H$-band measurements of $\eta$ Aql) we used the
452: shape of the radial velocity curve derived from the fit to the optical
453: data (i.e.\ by using the same Fourier coefficients); the IR data was
454: only used to determine an overall amplitude of the velocity curve. For
455: the IR points we used an effective p-factor of $1.41 \pm 0.03$ as
456: recommended by D. Sasselov (private communication) and based on an
457: analysis by \citet{sabbey95}, taking into account both the use of a
458: constant p-factor and the use of parabolic line fitting. The resulting
459: best-fit parameters are very similar to those based on optical radial
460: velocities (i.e. Table \ref{tab:result}): ${\rm D} = 333 \pm 30$ pc
461: and ${\rm R}=64.2 \pm 6 {\rm R}_{\odot}$. A similar fit for $\zeta$
462: Gem gives ${\rm D} = 359 \pm 37$ pc and ${\rm R}=62.2 \pm 5.7 {\rm
463: R}_{\odot}$. Hence we conclude that the effects of wavelength
464: dependence of the radial velocity are at present smaller than other
465: sources of uncertainty.
466:
467: \begin{table}
468: \begin{center}
469: \begin{tabular}{cccc}
470: & & & \\
471: \tableline
472: \tableline
473: Cepheid & Fit Type & Parameter & Best-Fit Results \\
474: & & & Value $\pm$ $\sigma_{Tot}$ ($\sigma_{Stat.}/\sigma_{Sys.}$) \\
475: \tableline
476: $\eta$ Aql & Pulsation Fit & Distance (D) & $ 320\pm32$ (24/21) pc \\
477: & No. Pts. = 18 & Radius (R) & $ 61.8\pm7.6$ (4.5/6.1) ${\rm R}_{\odot}$ \\
478: & $\chi^2_{dof}$ = 1.06 & Phase ($\phi$)& $ 0.02\pm0.011$ ($0.011/5 \times 10^{-4}$) cycles\\
479: \cline{2-4}
480: & Line Fit & $\theta_{UD}$ & $ 1.734\pm 0.070 (0.018/0.068)$ mas \\
481: & $\chi^2_{dof}$ = 13.4 & & \\
482:
483:
484:
485: \tableline
486: $\zeta$ Gem & Pulsation Fit & Distance (D) & $ 362\pm38$ (35/15) pc \\
487: &No. Pts. = 15 & Radius (R) & $ 66.7\pm7.2$ (6.3/3.4) ${\rm R}_{\odot}$ \\
488: &$\chi^2_{dof}$ = 1.82 & Phase ($\phi$) & $ 0.013\pm0.016$ ($0.016/3 \times 10^{-5}$) cycles \\
489: \cline{2-4}
490: & Line Fit & $\theta_{UD}$ & $ 1.613\pm 0.029 (0.017/0.024)$ mas \\
491: & $\chi^2_{dof}$ = 14.6 & & \\
492:
493: \tableline
494:
495: \end{tabular}
496: \caption{\label{tab:result} Best-fit Cepheid parameters and their
497: uncertainties, as well as mean apparent uniform-disk angular diameter
498: ($\theta_{UD}$) determined from fitting a line to all of the data. The
499: uncertainties of the best-fit parameters are broken down into
500: statistical ($\sigma_{Stat.}$) and systematic ($\sigma_{Sys.}$)
501: uncertainties. The goodness-of-fit parameter is a weighted $\chi^2$
502: divided by the number of degrees of freedom ($\chi^2_{dof}$) in the fit. The
503: $\chi^2_{dof}$ of the fits are calculated from data that does not have
504: the systematic (calibrator) uncertainty folded in since it applies
505: equally to all points. }
506:
507: \end{center}
508: \end{table}
509:
510:
511:
512: \begin{table}
513: \begin{center}
514: \begin{tabular}{clccc}
515: & & & & \\
516: \tableline
517: \tableline
518: Cepheid & Reference & Radius & Distance & Angular Diameter \\
519: & & ${\rm R}_{\odot}$ & (pc) & $\theta_{LD}$ (mas) \\
520: \tableline
521: \tableline
522: $\eta$ Aql & this work & $61.8\pm7.6$ & $ 320\pm32$ & $ 1.793\pm 0.070$ \\
523: &\cite{nordgren00} & & & $ 1.69\pm 0.04$ \\
524: &\citet{ripepi97} & $57\pm3$ & & \\
525: & \citet{esa97} & & $360^{+174}_{-89}$ & \\
526: &\citet{sasselov90}& $62\pm6$ & & \\
527: &\cite{fern89} & $53\pm5$ & $275 \pm 28$ & \\
528: &\citet{mb87} & $55\pm4$ & & \\
529: \tableline
530: $\zeta$ Gem & this work & $ 66.7\pm7.2$ & $ 362\pm38$ & $ 1.675\pm 0.029$ \\
531: & \citet{lane00} & $ 62\pm11$ & $ 336\pm44$ & $ 1.62\pm 0.3$ \\
532: & \citet{ker01} & & & $ 1.69^{+0.14}_{-0.16}$ \\
533: & \cite{nordgren00} & & & $ 1.55\pm 0.09$ \\
534: & \citet{esa97} & & $358^{+147}_{-81}$ & \\
535: & \citet{ripepi97} & $86\pm4$ & & \\
536: & \citet{bersier97} & $89.5\pm13$ & $498\pm84$ & \\
537: & \citet{ksn97} & $69.1^{+5.5}_{-4.8}$& & \\
538: & \citet{sabbey95} & $64.4 \pm 3.6$ & & \\
539: & \citet{mb87} & $65\pm12$ & & \\
540: \tableline
541: \end{tabular}
542: \caption{\label{tab:comparison} A comparison between the various available radius, distance and angular size
543: determinations. The \citet{nordgren00} results are based on $R$ band ($740$ nm) observations, while the
544: \citet{ker01} result is in the $K$ band ($2.2~\mu$m).
545: }
546: \end{center}
547: \end{table}
548:
549: The derived parameters (mean radius, distance and mean uniform-disk
550: angular diameter) can be compared to previously published values,
551: derived using a range of techniques (see Table \ref{tab:comparison}),
552: including parallax and a variety of surface brightness
553: techniques. There are also several interferometric diameter
554: measurements available in the literature, although to date no other
555: interferometers have directly resolved Cepheid pulsations. Thus,
556: directly measured angular diameters can only be compared in a
557: phase-averaged sense.
558:
559: \subsection{Surface Brightness Relations}
560:
561: A wide variety of Cepheid surface brightness relations have been used
562: by various authors \citep{be76,laney95,fg97} to derive Cepheid distance scales.
563: We define as surface brightness the quantity
564: \begin{equation}
565: \label{eqn:sb}
566: F_{i} = 4.2207 - 0.1 m_i - 0.5 \log(\theta_{LD})
567: \end{equation}
568: where $F_{i}$ is the surface brightness in magnitudes in
569: passband $i$, $m_i$ is the apparent magnitude in that band, and
570: $\theta_{LD}$ is the apparent angular diameter of the star. With the
571: above relation and a good estimate of $F_{i}$ one can determine
572: the angular diameter based on photometry alone. Conversely, given
573: measured angular diameters and multi-band photometry it is possible to
574: calibrate $F_{i}$ by finding a simple (e.g.\ linear) relation
575: between $F_{i}$ and a variety of color indices (e.g $V-K$). We
576: define the following relations
577: \begin{equation}
578: F_{V,1} = a + b (V-K)
579: \end{equation}
580: \begin{equation}
581: F_{V,2} = a + c (V-R)
582: \end{equation}
583: Note that
584: consistency requires a common zero-point (cf. an A0V star where
585: $(V-R)=(V-K)=0$).
586:
587: We used previously published $VRK$ photometry of $\eta$ Aql
588: \citep{barnes97} to derive its apparent magnitude in the above bands
589: as a function of phase by fitting a low-order Fourier series to the
590: published photometry, after first correcting for the effects of
591: reddening following the procedure outlined in \citet{evans93}. The
592: individual values of $E(B-V)$ were taken from \citet{fernie90}, and the
593: reddening corrections applied are listed in Table \ref{tab:red}. For
594: each diameter measurement we then used the Fourier series to derive
595: $m_V$ and $V-K$ at the epoch of observation, and using Eq. \ref{eqn:sb} we
596: derived the corresponding surface brightness. Results are shown in
597: Figure \ref{fig:fv} and listed in Table \ref{tab:sb}. We also
598: performed this type of fit using $\zeta$ Gem data. In this case we
599: used photometry from \citet{wj68} and \citet{mb84}.
600:
601:
602: \begin{table}
603: \begin{center}
604: \begin{tabular}{cccc}
605: & & & \\
606: \tableline
607: \tableline
608: Cepheid & $A_V$ & $A_R$ & $A_K$ \\
609: \tableline
610: \tableline
611: $\eta$ Aql & 0.515 & 0.377 & 0.055 \\
612: $\zeta$ Gem & 0.062 & 0.046 & 0.007 \\
613: \tableline
614: \end{tabular}
615: \caption{\label{tab:red} Reddening values used in deriving surface
616: brightness parameters for $\eta$ Aql and $\zeta$ Gem, based on
617: values of $E(B-V)$ from \citet{fernie90}. }
618: \end{center}
619: \end{table}
620:
621:
622: In Table \ref{tab:sb} we compare the derived surface brightness
623: relations to similar relations from work based on non-variable
624: supergiants \citep{fg97} and other Cepheid observations
625: \citep{nordgren01}. The $F_{V}$ vs. $V-R$ fits can also be compared with
626: the \citet{gieren88} result that the slope of the $V-R$ surface
627: brightness relation (c) is weakly dependent on pulsational period
628: ($P$) according to
629: \begin{equation}
630: c = -0.359 - 0.020 \log{P}
631: \end{equation}
632: which for $\eta$ Aql predicts $c = -0.376$ and for $\zeta$ Gem
633: $c = -0.379$.
634: These comparisons reveal generally good agreement between the various
635: relations in Table \ref{tab:sb}.
636:
637: \begin{table}
638: \begin{center}
639: \begin{tabular}{cccc}
640: & & & \\
641: \tableline
642: \tableline
643: Source & a & b & c \\
644: \tableline
645: $\eta$ Aql, this work &$3.941 \pm 0.005$&$-0.125\pm0.004$&$-0.375\pm0.002$\\
646: $\zeta$ Gem, this work &$3.946 \pm 0.011$&$-0.130\pm0.002$&$-0.378\pm0.003$\\
647: \citet{fg97} &$3.947 \pm 0.003$&$-0.131\pm0.003$&$-0.380\pm0.003$\\
648: \citet{nordgren00} &$3.941 \pm 0.004$&$-0.125\pm0.003$&$-0.368\pm0.007$\\
649: \tableline
650: \end{tabular}
651: \caption{\label{tab:sb} A comparison between the various surface brightness relations (see text for definitions).
652: }
653: \end{center}
654: \end{table}
655:
656:
657: \subsection{Period-Radius Relations}
658:
659: The relation between pulsational period and Cepheid radius has
660: received considerable attention in the literature, primarily because
661: early results based on different techniques were discrepant
662: \citep{fernie84,mb87}. Period-radius relations are also useful in that
663: they can indicate pulsation mode. This is important for calibrating
664: period-luminosity relations since different modes will yield different
665: relations \citep{fc97,nordgren01}.
666:
667: In Fig. \ref{fig:rad} we compare our measured Cepheid diameters to
668: the values predicted from a range of techniques:
669: \citet{bono98} calculate a period-radius relation from full-amplitude,
670: nonlinear, convective models for a range of metallicities and stellar
671: masses. \citet{gmb99} use the surface brightness technique based on $V$
672: and $V-R$ photometry and the \citet{fg97} result to derive radii for 116
673: Cepheids in the Galaxy and the Magellanic Clouds. They find an
674: intrinsic width in their relation of $\pm 0.03$ in log R.
675: \citet{laney95} also use the surface brightness technique for
676: estimating Cepheid diameters. However, they find that infrared
677: photometry ($K, J-K$) is less sensitive to the effects of gravity and
678: microturbulence (and presumably also reddening), and hence yields more
679: accurate results. For shorter periods ($\le 11.8$ days) their results
680: indicate smaller diameters as compared to other relations.
681:
682: Given the limited sample of only two radius measurements we can draw
683: only preliminary conclusions: (1) the general agreement between our
684: observations and the relations is good, and (2) the data seem to
685: prefer a shallower slope than e.g.\ the \citet{laney95} relation.
686: This latter observation will have to be confirmed with observations
687: of shorter-period Cepheids.
688:
689:
690: \section{Summary}
691:
692: We have measured the changes in angular diameter of two Cepheids,
693: $\eta$ Aql and $\zeta$ Gem, using PTI. When combined with previously
694: published radial velocity data we can derive the distance and mean
695: diameter to the Cepheids. We find $\eta$ Aql to be at a distance of
696: $320 \pm 32$ pc with a mean radius of $61.8 \pm 7.6 R_{\odot}$. We
697: find $\zeta$ Gem to be at a distance of $ 362\pm38$ pc, with a mean
698: radius of $ 66.7\pm7.2 R_{\odot}$, in good agreement with previous
699: work. The precision achieved is $\sim$ 10\% in the parameters; further
700: improvement is at present limited by our understanding of the details
701: of the Cepheid atmospheres. In particular the details of limb
702: darkening and projection factors need to be understood, with the
703: projection factors being the largest source of systematic
704: uncertainty.
705:
706: We note that these results do not rely on photometric surface
707: brightness relations, hence results derived here can be used to
708: calibrate such relations. We performed such calibrations and found
709: good agreement with previous results. We also note that at present we
710: have derived distances to only two Cepheids, and although the derived
711: distances are consistent with currently used period-luminosity
712: relations, it will be necessary to observe several more Cepheids with
713: this technique before worthwhile quantitative comparisons can be made.
714:
715: In the near future long-baseline interferometers will provide a great
716: deal of useful data in this area: in addition to further observations
717: of the brightest galactic Cepheids, the very long baselines currently
718: being commissioned at the Navy Prototype Optical Interferometer
719: \citep{arm01b} and the Center for High Angular Resolution Astronomy
720: array \citep{theo01} will allow direct measurements of the limb
721: darkening effects through observations of fringe visibilities past the
722: first visibility null. Given the close relation between limb darkening
723: and projection factors we expect that improvements in understanding
724: one will improve our understanding of the other. It is also clear
725: that additional photometry and radial velocity measurements would be
726: very useful. In particular $\zeta$ Gem suffers from a lack of good
727: infrared photometry, while concerns about level effects make infrared
728: radial velocity measurements like those of \citet{sasselov90} very
729: desirable.
730:
731:
732: \acknowledgements
733:
734: We thank D. Sasselov, A. F. Boden, M. M. Colavita, S. R. Kulkarni, and
735: R.R. Thompson for valuable comments. We also wish to thank K. Rykoski
736: for his excellent observational work. Observations with PTI are only
737: made possible through the efforts of the PTI collaboration, for which
738: we are grateful. Funding for the development of PTI was provided by
739: NASA under its TOPS (Toward Other Planetary Systems) and ASEPS
740: (Astronomical Studies of Extrasolar Planetary Systems) programs, and
741: from the JPL Director's Discretionary Fund. Ongoing funding has been
742: provided by NASA through its Origins Program and from the JPL
743: Directors Research and Development Fund. This work has made use of
744: software produced by the Interferometry Science Center at the
745: California Institute of Technology. This research has made use of the
746: SIMBAD database, operated at CDS, Strasbourg, France. B.F.L gratefully
747: acknowledges the support of NASA through the Michelson fellowship
748: program.
749:
750:
751:
752: \begin{thebibliography}{}
753:
754: \bibitem[Armstrong et al.(2001)]{arm01} \
755: Armstrong, J.~T., Nordgren, T.~E., Germain, M.~E., Hajian, A.~R., \
756: Hindsley, R.~B., Hummel, C.~A., Mozurkewich, D., \& Thessin, R.~N.\ 2001, \aj, 121, 476
757:
758: \bibitem[Armstrong et al.(2001b)]{arm01b} \
759: Armstrong, J.~T.~et al.\ 2001, American Astronomical Society Meeting, 198, 6303
760:
761: \bibitem[Albrow \& Cottrell (1994)]{albrow94} \
762: Albrow, M. D., Cottrell, P. L., 1994, \mnras, 267, 548--556
763:
764: %\bibitem[Alcock et al.(1995)]{alcock95}\
765: % Alcock, C. {\it et al.}, 1995, \aj, 109, 1653--1662
766:
767: %\bibitem[Andreasen(1988)]{andreasen88} \
768: % Andreasen, G. K., 1988, \aap, 196, 159--172
769:
770: \bibitem[Barnes \& Evans(1976)]{be76} Barnes, T.~G.~\& Evans, D.~S.\ 1976, \mnras, 174, 489.
771:
772: \bibitem[Barnes et al.(1997)]{barnes97} \
773: Barnes, T. G., Fernley, J. A., Frueh, M. L., Navas, J.G., Moffett, T. J., Skillen, I., 1997, \pasp, 109,645--658.
774:
775: \bibitem[Bersier et al.(1994)]{bersier94} \
776: Bersier, D., Burki, G., Mayor, M., Duquennoy, A., 1994, \aap, 108, 25--39
777:
778: %\bibitem[Bersier et al.(1996)]{bersier96} \
779: % Bersier, D., Burki, 1996, \aap, 306, 417--426
780:
781: \bibitem[Bersier (2002)]{bersier02} \
782: Bersier, D., submitted to ApJ.
783:
784: \bibitem[Bersier et al.(1997)]{bersier97} \
785: Bersier, D., Burki, G., Kurucz, R. L., 1997, \aap, 320, 228--236
786:
787: \bibitem[Boden et al.(1998) ]{boden98} \
788: Boden, A. F., van Belle, G. T.,
789: Colavita, M. M., Dumont, P. J., Gubler, J.,
790: Koresko, C. D., Kulkarni, S. R.,
791: Lane, B. F., Mobley, D. W., Shao, M.,
792: Wallace, J. K., 1998, \apjl, 505, L39
793:
794: \bibitem[Bohm-Vitense \& Proffitt(1985)]{bv85} \
795: Bohm-Vitense, E.~\& Proffitt, C.\ 1985, \apj, 296, 175--184.
796:
797: \bibitem[Bono, Caputo, \& Marconi(1998)]{bono98} \
798: Bono, G., Caputo, F., \& Marconi, M.\ 1998, \apjl, 497, L43
799:
800: \bibitem[ten Brummelaar et al.(2001)]{theo01} \
801: ten Brummelaar, T.~A.~et al.\ 2001, American Astronomical Society Meeting, 198, 6106
802:
803: \bibitem[Claret et al.(1995)]{claret95} \
804: Claret, A., Diaz-Cordoves, J., Gimenez, A., 1994, \aap SS, 114, 247--252.
805:
806: \bibitem[Cohen et al.(1999)]{cohen99} \
807: Cohen, M., Walker, R., Carter, B., Hammersley, P, Kidger, M.,
808: Noguchi, K., 1999, \aj, 117, 1864-1889.
809:
810: \bibitem[Colavita et al.(1999)]{colavita99} \
811: Colavita, M. M., Wallace, J. K., Hines, B. E., Gursel, Y., Malbet, F.,
812: Palmer, D. L., Pan, X. P., Shao, M. , Yu, J. W., Boden, A. F.,
813: Dumont, P. J., Gubler, J. Koresko, C. D., Kulkarni, S. R.,
814: Lane, B. F., Mobley, D. W., van Belle, G. T, 1999, \apj, 510 , 505--521.
815:
816: \bibitem[Colavita (1999)]{colavita99b} \
817: Colavita, M. M., 1999, \pasp, 111, 111--117.
818:
819: \bibitem[ESA(1997)]{esa97} \
820: ESA, 1997, The Hipparcos catalogue, SP-1200
821:
822: \bibitem[Evans \& Jiang(1993)]{evans93} \
823: Evans, N. R., Jiang, J. H., 1993, \aj, 106, 726--733.
824:
825: \bibitem[Feast(1999)]{feast99} \
826: Feast, M.\ 1999, \pasp, 111, 775
827:
828: \bibitem[Feast \& Catchpole(1997)]{fc97} \
829: Feast, M.~W.~\& Catchpole, R.~M.\ 1997, \mnras, 286, L1
830:
831: \bibitem[Fernie(1984)]{fernie84} \
832: Fernie, J.~D.\ 1984, \apj, 282, 641
833:
834: \bibitem[Fernie(1990)]{fernie90} \
835: Fernie, J. D., 1990, \apjs, 72, 153--162.
836:
837: \bibitem[Fernley, Skillen, \& Jameson(1989)]{fern89} \
838: Fernley, J.~A., Skillen, I., \& Jameson, R.~F.\ 1989, \mnras, 237, 947
839:
840: \bibitem[Fouque \& Gieren(1997)]{fg97} \
841: Fouque, P., Gieren, W.P., 1997, \aap, 320, 799-810.
842:
843: %\bibitem[Fry \& Carney (1997)]{fry97} \
844: % Fry, A. M., Carney, B. W., 1997, \aj, 113, 1073--1087.
845:
846: \bibitem[Gieren(1988)]{gieren88} \
847: Gieren, W.~P.\ 1988, \apj, 329, 790.
848:
849: %\bibitem[Gieren et al.(1989)]{gieren89} \
850: % Gieren, W. P., Barnes, T. G., Moffett, T. J., 1989, \apj, 342, 467--475
851:
852: \bibitem[Gieren, Moffett, \& Barnes(1999)]{gmb99} \
853: Gieren, W.~P., Moffett, T.~J., \& Barnes, T.~G.\ 1999, \apj, 512, 553
854:
855: \bibitem[Hindsley \& Bell(1986)]{hind86} \
856: Hindsley, R., Bell, R. A., 1986, \pasp, 98, 881--888
857:
858: \bibitem[Jacobsen \& Wallerstein(1981)]{jw81} \
859: Jacobsen, T.S., Wallerstein, G., 1981, \pasp, 93, 481--485.
860:
861: \bibitem[Jacobsen \& Wallerstein(1987)]{jw87} \
862: Jacobsen, T.S., Wallerstein, G., 1987, \pasp, 99, 138--140.
863:
864: \bibitem[Jacoby et al.(1992)]{jacoby92} \
865: Jacoby, G.~H.~et al.\ 1992, \pasp, 104, 599
866:
867: \bibitem[Kervella et al.(2001)]{ker01} \
868: Kervella, P., Coud{\' e} du Foresto, V., Perrin, G., Sch{\" o}ller, M., \
869: Traub, W.~A., \& Lacasse, M.~G.\ 2001, \aap, 367, 876
870:
871: \bibitem[Krockenberger, Sasselov, \& Noyes(1997)]{ksn97} \
872: Krockenberger, M., Sasselov, D.~D., \& Noyes, R.~W.\ 1997, \apj, 479, 875
873:
874: \bibitem[Lane et al.(2000)]{lane00} \
875: Lane, B. F., Kuchner, M. J., Boden, A. F., Creech-Eakman, M.,
876: Kulkarni, R. R., 2000, \nat, 407, 485--487.
877:
878: \bibitem[Laney \& Stobie (1995)]{laney95} \
879: Laney, C.D., Stobie, R.S., 1995, \mnras, 274, 337--360
880:
881: \bibitem[Moffett \& Barnes(1984)]{mb84} \
882: Moffett, T.~J.~\& Barnes, T.~G.\ 1984, \apjs, 55, 389--432.
883:
884: \bibitem[Moffett \& Barnes(1987)]{mb87} \
885: Moffett, T.~J.~\& Barnes, T.~J.\ 1987, \apj, 323, 280
886:
887: %\bibitem[McAlister {\it et al.} (1998)]{mac98}\
888: % McAlister, H.~A. {\it et al.}, 1998, {\em Proc. SPIE}, {\bf 3350}, 947--950.
889:
890: %\bibitem[Manduca (1979)]{manduca79} \
891: % Manduca, A., 1979, \aap SS, 36, 411--414
892:
893: %\bibitem[Mourard {\it et al.} (1997)]{mourard97} \
894: % Mourard, D. {\it et al.}, 1997, {\em A\&A}, {\bf 317}, 789--792.
895:
896: \bibitem[Nordgren et al.(2000)]{nordgren00} \
897: Nordgren, T., Armstrong, J.T., Germain, M.E., Hindsley, R.B., Haijan, A.R.,
898: Sudol, J.J., Hummel, C.A., 2000, \apj, 543, 972--978.
899:
900: \bibitem[Nordgren et al. (2001)]{nordgren01} \
901: Nordgren, T., Lane, B. F., Hindsley R. B., submitted to AJ.
902:
903: %\bibitem[Percy (1993)]{percy93} \
904: % Percy, J., 1993, \pasp, 105, 1422--1426.
905:
906: \bibitem[Press et al.(1986)]{press86}\
907: Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T.,
908: 1986, Numerical Recipies (New York: Cambridge Univ. Press).
909:
910: %\bibitem[Ridgeway et al.(1982)]{ridgeway82} \
911: % Ridgeway, S.T., Jacoby, G.H., Joyce, R.R., Siegel, M.J., Wells, D.C., \aj, 87, 680--684.
912:
913: \bibitem[Ripepi et al.(1997)] {ripepi97} \
914: Ripepi, V., Barone, F., Milano, L., Russo, G., 1997, \aap, 318, 797--804
915:
916: \bibitem[Sabbey et al.(1995)]{sabbey95} \
917: Sabbey, C.N., Sasselov, D.D., Fieldus, M.S., Lester, J.B., Venn, K.A., Butler, R.P.,
918: 1995, \apj, 446, 250--260
919:
920: \bibitem[Sasselov \& Lester(1990)]{sasselov90} \
921: Sasselov, D. D., Lester, J. B., 1990, \apj, 362, 333--345.
922:
923: \bibitem[Szabados (1991)]{szabados91} \
924: Szabados, L., 1991, Comm. Konkoly Obs., No. 96.
925:
926: \bibitem[Tanvir(1999)]{tan99} \
927: Tanvir, N.~R.\ 1999, ASSL Vol.~237: Post-Hipparcos cosmic candles , 17
928:
929: \bibitem[Welch(1994)]{welch94} \
930: Welch, D. L., 1994, \aj, 108, 1421--1426
931:
932: \bibitem[Wesselink(1946)]{wesselink46} \
933: Wesselink, A. J., 1946, BAN, 368, 91
934:
935: \bibitem[Wisniewski \& Johnson(1968)]{wj68} \
936: Wisniewski, W.~Z., \& Johnson, H.~L.\ 1968, Communications of the
937: Lunar and Planetary Laboratory, 7, 57
938:
939:
940:
941: \end{thebibliography}
942:
943:
944: \clearpage
945: \begin{figure}[ht]
946: \epsscale{0.8}
947: \plotone{f1a.eps}
948: \plotone{f1b.eps}
949: \caption[Cepheid Angular Diameters]{\label{diams} The angular diameters of $\eta$ Aql (top) and
950: $\zeta$ Gem (bottom) as a function of pulsational phase, together with
951: a model based on radial velocity data, but fitting for distance, mean
952: radius and phase shift. Also shown is the result of fitting a line to
953: all the data. The fits are extended past phase 0 for clarity.}
954: \end{figure}
955:
956: \clearpage
957: \begin{figure}[ht]
958: \epsscale{0.8}
959: \plotone{f2a.eps}
960: \plotone{f2b.eps}
961: \caption[Cepheid Surface Brightness]{\label{fig:fv} Dereddened ${\rm F}_V$ vs. $V-K$ (top) and
962: $V-R$ (bottom) for $\eta$ Aql. The solid line is the weighted linear
963: least-squares fit to the data. The dashed line represents the relation
964: from \citet{fg97}, and the dotted line represents the \citet{nordgren01}
965: result. }
966: \end{figure}
967:
968: \clearpage
969: \begin{figure}[ht]
970:
971: \plotone{f3.eps}
972: \caption[Perid-Radius Relations]{\label{fig:rad} Period-radius diagram for the two Cepheids $\eta$ Aql and $\zeta$ Gem, together with three relations available in the literature: a theoretical relation derived by \citet{bono98}, an optical surface brightness relation from \citet{gmb99} and an IR surface brightness relation from \citet{laney95}. }
973: \end{figure}
974:
975: \end{document}
976: