WorksPhysics and Astronomy
Towards a new model of the compact triple system Tauri I. Improved periods and preliminary elements of the triple system
Abstract
We analysed our new radial velocity measurements and photometric observations (covering a time interval of 46838~days and 39100~days, respectively) as well as those from previous studies to derive accurate ephemerides for both orbits of the compact triple system Tau. In the case of the inner eclipsing subsystem, we obtained a~linear ephemeris of T min.I=HJD 2440998.86069(35)+3.952952845(81)xE, improving the accuracy of the previous solutions by more than an order of magnitude. This allowed us to reproduce well all available light curve minima, including the first one recorded in 1844. Furthermore, the period of the outer subsystem, determined from radial velocity analysis, appears to be constant over the interval of available observations, with the corresponding linear ephemeris being T periast.=HJD 2456822.5(3.6)+33.02480(28)xE. Assuming near coplanarity of the inner and outer orbits, we estimated the preliminary physical properties of the whole system. We found that the secondary of the inner subsystem is filling or nearly filling its Roche lobe. We used the long-term stability of the eclipsing subsystem period to place a strong upper limit on the rate of the proposed episodic mass transfer to its primary component. Our results constitute a preparatory step for a follow-up study that will also include an analysis of a rich collection of optical spectro-interferometric observations and use a full-fledged numerical model to describe the evolution of the orbital architecture of the system. Particular attention will be focused on determining the inner orbit forced eccentricity and understanding the absence of the apsidal advance of the outer orbit in our present solution.
Corresponding author: [email protected]
Introduction
Compact triple systems are extremely interesting objects in stellar astrophysics, as they represent unique laboratories to test various hypotheses related to their origin, they are subject to a plethora of complicated evolutionary pathways, and they offer a possibility of revealing traces of dynamical effects of various timescales within affordable decade-long periods. The latter, if properly modelled, consequently provides a unique tool to constrain the physical parameters of the stars. Analysis of the data from Kepler and the Transiting Exoplanet Survey Satellite (TESS) space-borne surveys has provided spectacular examples of systems in this category. For example, Kostov et al. [31] reported the discovery of the so-far most compact triply eclipsing triple system, TIC 290061484 = 2MASS J21042626+5113320, with an outer period of only, and Borkovits et al. [9] reported ten new compact systems with outer periods between and . Even more recently, Rappaport et al. [42] discovered TIC 366240660, a triply eclipsing system with an outer orbit period of in a rigorous 13:1 mean motion resonant configuration with the inner orbit. Although these new systems are fascinating in many respects and have brought evidence of previously unseen configurations, Tau – with an outer orbit period of only 33 d – remains the prime and archetypal representative of a stellar compact triple hierarchy. Among other reasons, its uniqueness stems from its brightness and its long observational history. Considering the availability of numerous spectroscopic and photometric observations of Tau (briefly overviewed in Sect. 2) and reporting results from our own dedicated campaign, which provide a new dense series of spectroscopic and photometric observations, the goal of the present paper is to provide reliable and accurate ephemerides for both inner and outer orbits in the Tau system. These will serve as a starting point for a follow-up study in which we will make use of additional interferometric observations and construct a state-of-the-art dynamical model of this remarkable triple star.
Relevant observational facts about Tau
The star Tau (HD 25204, HR 1239, HIP 18724) is one of the brightest eclipsing binaries in the northern sky (, , ; [29]). Its light variations were reported as early as 1848 by Baxendell [1], who already realised that the period of light changes must be close to 4 days. Later, it turned out that the very first minimum was in fact already observed by Schmidt [46] on January 21, 1844. The first photoelectric light curve was published by Stebbins [50], who observed the star relative to Tau, Ori, Ori, Tau, and Tau. He discovered (see also Stebbins [51]) that Ori is an ellipsoidal variable with a period of and a full amplitude of . Using a green-sensitive diode, Stebbins observed Tau through an attenuating filter, carefully calibrated its transparency, compensated also for the variations of Ori, and finally reduced all observations relative to Tau. Interestingly enough, Tau is now also known to be an eclipsing binary with a period ([5, 35, 36, 11]). Stebbins’s observations defined the eclipsing light curve of Tau quite well and allowed him to derive the ephemeris for the primary minimum:
Another (blue) light curve of Tau was obtained by Nikonov [37] relative to Tau and Tau. The first light curve of Tau was obtained by Grant [20] relative to HR 1194. His comparison star also turned out to be a light variable (of CP2 type) with a period of and a range of in Strömgren ([43, 57]) and is now known as V766 Tau.
The changes in radial velocity (RV) of Tau were first reported by Belopolsky [2], and the first RV curve with a period of , based on 23 Potsdam spectra, was published by von Aretin [55]. The presence of a close tertiary moving around the centre of mass of the eclipsing binary with a period of was discovered by Schlesinger (1911b). McLaughlin [34] obtained a new series of spectra and concluded that a better value for the long period is . Ebbighausen & Struve [15] obtained an additional series of spectrograms at several observatories and analysed them together with previously published ones to arrive at a period of for the long orbit. They noted that the Stebbins [50] ephemeris (1) for the short orbit satisfies all available RVs. Later, Grant [20] studied 20 coude spectra from the McDonald 82-inch reflector and demonstrated the presence of the weak lines of the secondary in the H and Mg II 4481 Å lines. He did not publish any RVs but estimated the component masses to be 6.06 and 1.61 M. Casini et al. [12] measured RVs on 16 photographic grating spectra with a dispersion of 34 Å mm and derived new orbital elements. However, they did not publish the individual RVs. Fekel & Tomkin [16] obtained the first electronic (Reticon) spectra of Tau covering the wavelength region from 4460 to 4515 Å, and they were able to detect the signal of the secondary in several metallic lines and measure their RVs. They found that the RV residuals from the orbit follow the orbit for both primary and secondary RVs. They also provided some arguments in favour of near coplanarity of both orbits. Berdyugin et al. [3] obtained 69 polarimetric observations of Tau between December 2012 and November 2016 at two observatories. They found small periodic changes with double-wave phase curves in the normalised Stokes parameters and in all three photometric passbands. Using an analytical method, they estimated the orbital inclination of the inner eclipsing binary to be between about and , while for scattering models, they found values between and . They confirmed that the inner system is a semi-detached one, with the cooler secondary filling its Roche lobe. They also found that the revolution of the inner system, as seen on the plane of the sky, is clockwise.
Observational material used
Throughout this paper, we express the times of observations in the reduced Julian date defined as
with the exception of very old photometry obtained before JD 2400000. It is true that it would be better to use the barycentric Julian dates (BJDs) consistently with all effects of non-uniform Earth rotation included. However, for many of the datasets we used, the accurate times of mid-exposures are not known, and only Julian dates to three decimal digits are provided. Since the differences between HJDs and BJDs in days are only in the fourth or fifth decimal place, it is impossible to calculate BJDs precisely enough for such data. Doing so is only possible for more recent datasets, and we postpone this analysis to the follow-up study.
Spectroscopy
We compiled and carefully re-examined all published RV measurements of Tau. Their log is provided in Table 1 and details on the individual datasets are in Appendix A. Our new observational material consists of the following series of electronic spectra from five observatories secured with seven different spectrographs: Red segments of 25 usable ESO La Silla Coralie CCD spectra; ESO La Silla FEROS echelle CCD spectra, one from 2008 and eleven from a close series in 2013; ESO La Silla HARPS echelle CCD spectra; Ondřejov 64 red and five blue CCD spectra; Lisboa 0.356 m Littrow CCD spectra; CHIRON echelle spectra from Cerro Tololo Inter-American Observatory (CTIO); and echelle spectra from the Skalnaté Pleso Observatory. A complete log of these electronic spectra is given in Table 2 and the details on the spectra reduction and RV measurements can also be found in Appendix A.
| Spg. No. | Time interval (RJD) | No. of RVs prim./sec. | Source |
| 1 | 14230.47–14268.38 | 5/2 | A |
| 2 | 15689.50–15849.27 | 23/0 | B |
| 3 | 17830.82–20210.51 | 28/0 | C |
| 4 | 17523.77–20915.53 | 107/0 | D |
| 5 | 23743.70–24157.67 | 91/0 | E |
| 6 | 28048.01–29295.62 | 36/0 | D |
| 7 | 34326.75–35143.78 | 68/0 | D |
| 8 | 35017.01–35036.80 | 2/0 | D |
| 9 | 44922.72 | 1/0 | F |
| 10 | 44177.90–44955.89 | 44/44 | G |
Table 1. Published available RV observations of Tau.
| Spg. No. | Time interval (RJD) | No. of sp. | Wavelength range (Å) | Spectral res. |
| 11 | 54741.80–55501.89 | 25 | 6320–6710 | 60000 |
| 12 | 54807.56–56564.75 | 12 | 3700–9220 | 48000 |
| 13 | 55060.87–56578.90 | 17 | 6260–6910 | 115000 |
| 14 | 56154.62–58390.53 | 64 | 6260–6740 | 11700 |
| 14 | 56235.36–57075.25 | 5 | 4270–4510 | 11700 |
| 15 | 57354.38–58075.39 | 17 | 6520–6700 | 14000 |
| 16 | 59472.87–59504.85 | 165 | 4500–7600 | 79000 |
| 16 | 59865.70–59911.76 | 250 | 4500–7600 | 79000 |
| 17 | 60622.57–60729.34 | 13 | 4240–7260 | 30000 |
| 17 | 60920.58–61068.40 | 20 | 4240–7260 | 30000 |
Table 2. Available electronic spectra of Tau.
Photometry
We carefully compiled all available published photometric observations with known times of observations and combined them with new photoelectric observations secured at the Hvar Observatory and with observations secured in the private observatory of Robert Uhlář. The log of all available photometric observations is in Table 3, and the details on photometric datasets and their homogenization and reductions can be found in Appendix § 10. Inspecting the phase plots, we found that the data from files A, F, and K have higher errors than all remaining Notes. Spectrographs and telescopes in column “Spg. No.”:
| File | Time interval (HJD) | No. of obs. | Passbands | HD of comparison / check star | Source |
| A | 2396338.4–2396354.4 | 5 | visual | 25490/many | Baxendell [1] |
| B | 2421159.8–2421635.5 | 150/145 | pe-cell | 21364 | Stebbins [50] |
| B | 2421164.9–2421321.6 | 24 | pe-cell | 31237 | Stebbins [50] |
| B | 2421202.7–2421296.5 | 16/15 | pe-cell | 30836 | Stebbins [50] |
| B | 2421482.9–2421635.6 | 56/55 | pe-cell | 23793 | Stebbins [50] |
| B | 2421514.9–2421670.6 | 21 | pe-cell | 26912 | Stebbins [50] |
| C | 2428057.5–2428115.6 | 48 | blue pe | 26912/25490 | Nikonov [37] |
| C | 2428421.5–2428484.4 | 46 | blue pe | 26912/25490 | Nikonov [37] |
| C | 2428788.5–2428816.6 | 38 | blue pe | 26912/25490 | Nikonov [37] |
| D | 2435091.6–2435160.7 | 228/230/310 | UBV | 24155/26171 | Grant [20] |
| E | 2447911.7–2448828.7 | 94 | – | Perryman & ESA [38] | |
| F | 2454216.1–2454232.3 | 481 | STEREO | – | this paper |
| F | 2454251.5–2454273.9 | 497 | STEREO | – | this paper |
| F | 2454559.9–2454576.4 | 525 | STEREO | – | this paper |
| F | 2454642.7–2454661.8 | 464 | STEREO | – | this paper |
| F | 2454903.7–2454922.3 | 561 | STEREO | – | this paper |
| F | 2455028.8–2455051.0 | 491 | STEREO | – | this paper |
| F | 2455247.6–2455266.8 | 632 | STEREO | – | this paper |
| F | 2455417.7–2455436.1 | 505 | STEREO | – | this paper |
| F | 2455592.1–2455610.4 | 571 | STEREO | – | this paper |
| F | 2455813.1–2455829.4 | 421 | STEREO | – | this paper |
| F | 2455939.4–2455955.7 | 512 | STEREO | – | this paper |
| F | 2456200.7–2456217.1 | 362 | STEREO | – | this paper |
| G | 2456523.6–2460583.5 | 226 | UBVR | 25330/26171 | this paper |
| H | 2456590.4–2456727.4 | 310/435/437/441 | BVRI | 25330 | this paper |
| I | 2457732.2–2457746.9 | 4314 | MOST | – | this paper |
| I | 2458059.3–2458093.5 | 17943 | MOST | – | this paper |
| J | 2458438.1–2458463.7 | 1214*) | TESS | – | this paper |
| K | 2458807.1–2459192.3 | 2426 | RGB | all-sky | Yin et al. [58] and priv.com. |
| J | 2459174.2–2459200.2 | 1248*) | TESS | – | this paper |
| J | 2459447.7–2459524.4 | 3494*) | TESS | – | this paper |
| J | 2460208.6–2460259.9 | 2422*) | TESS | – | this paper |
Notes. *)Normal data points over 0d.02 to reduce the excessive number of observations
Table 3. Available photometric observations of Tau.
1... Pulkowa 0.75 m refractor, 2-prism spg. 14 Å mm;
2... Potsdam 0.80 m refractor, prism spg., 27 Å mm at H, modern wavelengths;
3... Allegheny 0.79 m reflector, 1-prism spg., 40 Å mm, old wavelengths;
4... Allegheny 0.79 m reflector, 1-prism spg., 40 Å mm, modern wavelengths;
5... Ann Arbor 0.95 m reflector, 1-prism spg., 40 Å mm at H;
6... Dominion Astrophysical Observatory 1.83 m reflector, prism spg.;
7... MtWilson 1.52 m reflector, prism. spg., 36 Å mm at H;
8...Lick 0.91 m refractor, Mills 3-prism spg.;
9...Kitt Peak Observatory coudé auxiliary 0.90 m reflector, grating spg. 16.9 Å mm;
10... McDonald 2.7 m reflector, grating. Reticon spg.;
Abbreviations in column “Source” stand for:
A... Belopolsky [2]; B... von Aretin [55]; C... Schlesinger [45]; D... Ebbighausen & Struve [15]; E... McLaughlin [34]; F... Zentelis [59]; G... Fekel & Tomkin [16];
datasets, and we excluded them from our analyses. However, as shown in Fig. B.1, even these data match the accurate solution found in the next section.
PHOEBE solution for all good photometric observations
Since we succeeded in accumulating and homogenizing a very representative collection of photometric observations covering an interval of 39100 days, we first decided to verify the stability of the orbital period of the inner system. To do this, we used the programme PHOEBE 1 (Prša & Zwitter 2005, 2006), which works quickly and is quite sufficient for the given purpose. The inner binary was reported to be a semi-detached system, but as we mention later, there is a suspicion that the mass transfer process can be non-continuous and that the Roche-lobe overflow occurs only near the periastron passages of the outer system. For this reason, we treated the binary as ‘detached’ during our solution. Nevertheless, it did turn out that the secondary is in or very close to the state of Roche lobe filling. This is illustrated by the light curve model mesh plots shown in Figure 1 and by the fact that the side radius of the secondary from the PHOEBE solution corresponds to that of the Roche lobe for the obtained mass ratio.
Figure 1. Top: Mesh plot showing the eclipsing binary system at elongation with the primary approaching. The position of the Lagrangian point is shown. Bottom: Mesh plot at the phase of a secondary mid-eclipse.
During modelling, we kept the effective temperature of the primary fixed at 23000 K, which corresponds to its spectrum.
Notes. Spectrographs and telescopes in column “Spg. No.”:
11... ESO LaSilla Leonard Euler Nasmyth 1.2 m reflector, Coralie echelle CCD spectrograph;
12... ESO LaSilla 2.2 m Cassegrain reflector, FEROS echelle CCD spectrograph;
13... ESO LaSilla 3.6 m reflector, HARPS echelle CCD spectrograph;
14... Ondřejov 2.0 m reflector, coudé grating SITe-5 CCD spg.;
15... Lisboa IGeO 0.356 m SC reflector, Littrow LHIRES III spg.;
16... CTIO 1.5 m reflector, CHIRON echelle spectrograph;
17... Skalnaté Pleso 1.3 m Nasmyth-Cassegrain reflector, echelle spectrograph.
Our solution indicates that the orbital period is very stable during the whole time interval of available accurate observations, as seen in the phase plots of all available light curves in Figure 2. The corresponding linear ephemeris is

Figure 2. Top: Light curves based on all available good photometries for the final PHOEBE solution. Bottom: Interval of phases near the primary minimum.
The secular constancy of the orbital period can also be illustrated quantitatively by noting that the new ephemeris (2) predicts that the epoch of the primary minimum derived by Stebbins [50] is RJD 21506.85021, which is in perfect agreement with the ephemeris (1). We also note that the very first minimum was observed at Hamburg on January 21, 1844, at 7.334 hours UT. If we assume that the local Hamburg time of 8 hours was the one used by Schmidt [46], namely HJD 2394587.3082, the ephemeris (2) would situate it at the orbital phase 0.017. Finally – as illustrated by Fig. B.1 – the less accurate STEREO and Tibet Ali Observatory photometric observations are also fitted very well by our new ephemeris. We therefore conclude that the orbital period of the inner binary has been constant over the past 180 years of observations (within the uncertainties stated in Eq. (2)).
Preliminary orbital solutions
In order to make use of all previously published RVs, we used the programme reSPEFO2 (see the appendix in Harmanec et al. [27]) to determine the RVs from our new spectra, and we combined the two sets of RVs into a single dataset. Our RV values represent the mean values measured using He I and metallic lines. The Balmer lines were used to derive RVs for the primary component only. This is because the position of the secondary line is always strongly biased by being blended with the wide and steep absorption wings of the primary line.
Next, we used the programme FOTEL (Hadrava [21, 24]) to obtain the preliminary orbital solutions. We found this to be an adequate initial step since FOTEL provides a simultaneous solution for the inner and outer orbits. Nevertheless, there are two simplifying assumptions we made: (i) We fixed the eccentricity of the inner orbit to zero, and (ii) we considered the orbital parameters of the inner and outer orbits to be constant (see Sect. 8.4 for more discussion). We also kept the period and the epoch of the primary mid-eclipse fixed at the values derived with PHOEBE 1. After several trial solutions, demonstrating good agreement between separate solutions based on either only primary or only secondary RVs, we determined the final solution based on all primary and secondary RVs. To infer possible systematic effects of the measurements and to estimate rms errors separately for the primary and secondary RVs, respectively, an independent value of the systemic velocity was used for the two sets of RVs. The orbital elements are provided in Table 4, and the corresponding RV curves are shown in Figure 3 and Figure 4, each folded to the phase of the corresponding orbital period.

Figure 3. Preliminary RV curves based on mean RVs from the literature and on ours measured in reSPEFO2. In the process of the triple-star solution using the FOTEL program, the RVs were pre-whitened for the orbit. Black circles denote the primary RVs, open circles the secondary RVs. The residuals from the global fit are shown by blue dots for the primary, and red dots are for the secondary.
Figure 4. Preliminary average RV curve of the motion of the close pair in the 33.025 orbit with the unseen tertiary, pre-whitened for the motion in the inner orbit and based on the mean RVs. Top: RV curve based on the residuals of the primary RVs for the short orbit. Bottom: Noisier RV curve based on the residuals of the secondary RVs.
| Element | Value |
| (d) | 3.952952845 fixed |
| (RJD) | 40998.86069 fixed |
| 0.0 fixed | |
| (km s) | 17.08 0.20 |
| (km s) | 16.48 0.50 |
| (km s) | 55.51 0.28 |
| 0.2712 0.0017 | |
| (km s) | 205.02 |
| (d) | 33.02480 0.00028 |
| (RJD) | 56822.5 3.6 |
| (RJD) | 56830.3 |
| 0.054 0.033 | |
| (°) | 269 39 |
| 0.1927 0.0060 | |
| (km s) | 10.70 |
| () | 5.677 |
| () | 1.540 |
| () | 20.33 |
| No. of RVs (prim./sec.) | 993/499 |
| (km s) | 6.14 |
| (km s) | 10.95 |
Table 4. Orbital elements based on all published RVs and on mean reSPEFO2 RVs for both binary components of the eclipsing subsystem.
Spectra disentangling with KOREL
To verify the results of standard RV measurements, we analysed all available electronic spectra with a more sophisticated approach, namely spectra disentangling. First, we derived trial solutions in which the spectra were split into several subsets according to their time of observations. Individual subsets covered time intervals no longer than 1 year. We separately disentangled every subset that contained at least ten spectra using the programme KOREL (Hadrava [22, 23, 25]). All trials and final calculations were carried out by Jana Švrčková with the help of the VO-KOREL service (see Škoda et al. [56]). Since the results were consistent across seasons and different spectral lines, in the end we disentangled all available spectra at once in spectral segments centred on the Si II 6347 and 6371 Å doublet, on the H line, and on He I 6678 Å (see Figure 5) to obtain the solutions given in Table 5. The practical reason for this choice was that these spectral segments were available for almost all spectra we used. Not surprisingly, there is no trace of the secondary star in the He I line, while both components were disentangled in the H and Si II lines.
| Element | H | Si II | He 6678 | red |
| (d) | \multicolumn{4}{c}{3.952952845 – fixed} | |||
| (RJD) | \multicolumn{4}{c}{40998.86069 – fixed} | |||
| \multicolumn{4}{c}{0.0 – fixed} | ||||
| (km s) | 57.87 | 56.28 | 59.25 | 56.59 |
| 0.278 | 0.276 | — | 0.273 | |
| (km s) | 208.253 | 203.670 | — | 207.273 |
| (d) | 33.02738 | 33.02750 | 33.01802 | 33.02975 |
| (RJD) | 56822.62 | 56821.89 | 56822.13 | 56821.78 |
| 0.200 | 0.149 | 0.102 | 0.154 | |
| (°) | 274.2 | 266.6 | 265.4 | 269.22 |
| (km s) | 9.71 | 10.51 | 10.63 | 9.95 |
Table 5. KOREL solutions for the three spectral segments and a solution (“red”) based on all three spectral segments combined.
Figure 5. Normalised disentangled line profiles. Bottom profiles correspond to the primary component. Those from the secondary have been shifted upwards for a constant in the relative flux.
The KOREL solution of Table 5 agrees remarkably well with the overall FOTEL solution of Table 4 and also with the previous result obtained by Fekel & Tomkin [16]. Nevertheless, we point out two issues of tension in the two results. First, the solution of FOTEL leads to a considerably lower eccentricity of the outer orbit than KOREL. Second, there appears to be no apsidal precession of the outer orbit over the 18 years spanned by the spectral observations in the KOREL solution (even though the software would provide a linear drift in this orbital parameter if
it resulted in a better fit of the data; see also Sect. 8.4). We believe the two problems are related and will be resolved by using a full-fledged dynamical model of the system in the planned forthcoming study. On the positive side, we note that the epochs of the periastron passage of the outer orbit agree well for FOTEL and all KOREL solutions.
Preliminary system properties
By adopting the KOREL fit to the whole set of available spectra as a nominal solution, we obtained the following values for the inner subsystem: , , and . For the inclination of and the relative side radii of the PHOEBE 1 solution, this implies quite plausible values of and , a semi-major axis of , and component radii of and . In contrast to the result of [16] we found that the Roche-lobe secondary is slightly larger than the primary. For the outer subsystem, adopting again the global KOREL solution with km s, , the same system inclination, and the total mass of the inner binary of , we obtained , , and a separation at periastron of . The mass of the third body would correspond to a K4 star, according to the tabulation by [26].
Returning to the remarkable stability of the eclipsing binary period () for over more than a century (Eq. (2)), we can esti-
Notes. Separate systemic velocities were allowed for the primary and secondary. A triple-star solution was derived with the program FOTEL.
mate an upper limit for the hypothetical mass transfer from the secondary to the primary. Assuming for simplicity an adiabatic transfer conserving total mass and angular momentum of the system, the rate of mass loss from the secondary produces a period change whose values are related by
where (see, for example, Eq. 31 in Sepinsky et al. [47]). Having found that (orbital period and its rms error) over a time interval , the upper limit of the period change is . Using this estimate in Eq. (3), we obtained the following:
Adopting the values and from Eq. (2), yr and the total mass estimated above, we
have . This is a very low but reasonable value. We note that in semi-detached systems with the detected change in orbital period, the rates of mass transfer are usually of the order of or higher (see, e.g., Çelik & Kahraman Aliçavuş [13]). Despite a non-detection of the orbital period change for Tau, the upper limit on magnitude of the mass-transfer rate may be useful if this process had to be connected with photometric micro-variability (see Sect. 8.2 and Berdyugin et al. [3]). With the orbital solution caveats discussed above, which could also influence the inferred values of physical parameters to some degree, we consider the latter to be a valuable initial estimate for more in-depth modelling in the future.
Some peculiar phenomena worth of further investigation
In this section we briefly highlight a couple of puzzling phenomena that we noticed in our analysis. We believe that they are just different manifestations of the fact that Tau is a highly dynamical system down to the shortest timescale of the orbital period of the eclipsing binary.
Occasional line-core asymmetries or even line doubling for the primary component
While measuring the RVs interactively in reSPEFO2, we noted that especially the He I lines and the Mg II 4481 Å line of the primary component exhibit either blue or red asymmetries in the line cores and even the line doubling. Two outstanding examples of such variations are shown in Figure 6. They do not seem to correlate with any particular orbital phase of the close eclipsing subsystem, and they were observed far from the epochs of the periastron passage in the outer orbit. They might indicate some temporal changes in the large-scale mass transfer between the binary components.
Figure 6. Two peculiar HARPS spectra in the neighbourhood of the He I 4471 Å and Mg II 4481 Å lines.
Physical microvariability of the brightness
During inspection of the TESS and MOST light curves, we observed that some seemingly complicated microvariability is present. As an example, we show the residuals of the PHOEBE 1 light curve fit for the long series of TESS observations in segments 42, 43, and 44 in Figure 7. In this context, we note that Berdyugin et al. [3] analysed non-periodic phenomena in their polarisation observations, which they attributed to sporadic changes in the mass transfer rate from the secondary to the primary of the eclipsing subsystem. It is tempting to connect them with the photometric microvariability shown in Figure 7. The limit on the mass-transfer rate derived above may be a useful quantitative constraint for further modelling.
Figure 7. residuals from the PHOEBE 1 light curve fit of accurate TESS photometry from segments 42, 43, and 44 versus time.
Eclipse timing variations
The very accurate photometry provided by the Kepler and TESS surveys and their discoveries of very compact triples and higher stellar hierarchies enabled analysis and scientific use of eclipse timing variations (ETVs; see, e.g., Borkovits [6]). Basically, ETVs are differences between the epochs of the primary and secondary light curve minima predicted by the linear ephemeris (2) and their true instants. In compact systems such as Tau, ETVs are primarily caused by gravitational interaction between the inner and outer orbits (often referred to as physical ETVs) overshadowing the traditional light travel time effect [6].
Although the ETV analysis proved very useful in the studies of many systems, here a complication stemed from the microvariability of the photometric signal well documented away from the eclipse epoch (Sect. 8.2). This caused a perturbation propagating towards the eclipses, making the photometric signal near the minima asymmetric and time variable. The inferred ETV series may thus contain a non-negligible noise component in the case of Tau.
With this caveat in mind, we gave a preliminary look at the ETVs derived from all available TESS and MOST data (seven sectors of TESS and two seasons of MOST; Table 3). We used three different methods (fitting the light curve template, the classical Kwee-van Woerden method, and the polynomial fitting method) to derive mid-eclipse epochs for all primary and secondary eclipses. The final ETV data, representing the robust mean of the three individual values, are shown in Figure 8. Since the principal physical ETV signal has the period of the outer orbit in the system, we used the phase of the outer orbit at the abscissa instead of time in Figure 8. The fundamental quadrupole interaction ETV signal has a predicted amplitude (see, e.g., Eqs. (20) and (5) in Borkovits et al. [7, 8], respectively):
Figure 8. Eclipse timing variation diagram of the times of the mid-eclipse of the inner orbit, plotted versus phase of the outer orbit from Table 4. Larger red dots denote the primary, and smaller blue dots indicate the secondary eclipses. The red arrow indicates the instant of the periastron passage, while the blue one represents the time of maximum RV.
where we used for the outer orbit eccentricity. However, additional contributions will certainly arise due to a significant octupole orbital interaction (due to non-equal masses and of the binary) and apsidal and tidal contributions. In particular, the apsidal contribution depends sensitively on the mean eccentricity of the inner orbit. In this paper, we fixed its value to zero, but in Sect. 8.4 we argue that a small mean value of up to should be forced by the gravitational perturbation from the third component in the system. If so, the apsidal contribution to ETVs is of the order of , a few thousand a day (e.g., Giménez & Bastero [19]). At this moment we can conclude that the signal shown in Figure 8 has the expected order of magnitude, has the principal phasing with the period of the outer orbit, and, most importantly) is not dominated by noise. We plan to make full use of these ETV series in our forthcoming study.
Puzzles of the preliminary orbital solution of Tau
Given the compactness of the hierarchical triple system Tau, we expect it to evolve highly dynamically (see, e.g. Borkovits [6]). In this respect, the absence of apsidal motion of the outer orbit appears to be a curious element of our results (Sect. 6). Indeed, adopting a near-coplanar geometry of the system, a simple estimate of the outer orbit apsidal rate from the quadrupole interaction of the inner and outer orbits would be
where is the mean motion of the outer orbit; see, for example, Eq. (36) in Soderhjelm [49] or Eq. (86) in Breiter & Vokrouhlický [10]. With all the parameters on the right-hand side of (6) known, we would expect to circulate with a period of years. A direct numerical simulation, which also contains higher-order orbital couplings, confirms this value within an uncertainty of . Either our analysis suffers from oversimplifications that propagate principally to the solution of outer orbit periapse behaviour or some unexpected dynamical processes have to be included in the future solution.
As far as the first possibility is concerned, the main issue may be related to our assumption of zero eccentricity, , of the inner orbit. However, the significant difference in masses and triggers an octupole interaction between the inner and outer orbits, which should force a small but non-zero binary eccentricity. Our numerical tests as well as previous analytical estimates [e.g. 17, 18] provide a mean value of for the inner eccentricity. This value is small and may not be easily detectable in the available photometry or spectroscopic observations directly. However, a cumulated effect of the inner orbit apsidal precession could provide more hints from the analysis of the ETVs of the eclipsing binary. Our preliminary data shown in Figure 8 still leave room for apsidal contribution, but they need to be analysed using a more complex approach.
Data availability
Tables A.1 and B.1 – B.8 are available only in electronic form at the CDS via anonymous ftp to cdarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/ .
Acknowledgements. We acknowledge the use of the programme PHOEBE developed by Andrej Prša, the programmes FOTEL and KOREL, written by Petr Hadrava, and the programme reSPEFO2 by Adam Harmanec. Some of the Ondřejov spectra were obtained by Jana Nemravová - Benešová (who left astronomical research in the meantime) and also by the late Pavel Mayer. The authors also thank Jia Yin, who kindly provided us with his all-sky RGB photometry from Ali Observatory in Tibet, and Danielle Bewster for advice on STEREO photometry. Based on new spectroscopic observations from the following observatories: Cerro Tololo Inter-American Observatory, ESO LaSilla, Ondřejov, Stará Lesná, and Lisboa, and on new photometry from Hvar, Hipparcos, STEREO, and Ali Observatory, and also on the very last photometric observations carried out in 2016 and 2017 by the memorable Canadian space observatory MOST and downloaded through the Vienna tracking station, often with a heroic effort by the staff there. The paper also includes data collected by the TESS mission, which are publicly available from the Mikulski Archive for Space Telescopes (MAST). Funding for the TESS mission is provided by NASA’s Science Mission Directorate. Over the years, the research of the Czech authors was supported by the grants 205/06/0304, P09/10/0715, and GA15-02112S of the Czech Science Foundation, 678212 and 250015 of the Charles University Grant Agency research project AV0Z10030501 of the Academy of Sciences of the Czech Republic, Research Programme MSM0021620860 Physical study of objects and processes in the solar system and in astrophysics of the Ministry of Education of the Czech Republic. Hrvoje Božić obtained financial support from the Croatian Science Foundation under the project 6212 “Solar and Stellar Variability”. The research of Theo Pribulla was supported by the VEGA grant of the Slovak Academy of Sciences No. 2/0033/26 and by the Slovak Research and Development Agency under contracts Nos. APVV-24-0160 and APVV-25-0583. Constructive suggestions from an anonymous referee helped improve the clarity of the paper. The authors acknowledge the use of the electronic bibliography maintained by the NASA/ADS system and by the CDS in Strasbourg.
Overview of available spectroscopic observations
The initial reductions of all electronic spectra were performed using the IRAF and SPEFO programmes.4 Spectra normalization, and cleaning from cosmics and flaws were carried out with the Java program reSPEFO2 (see the appendix in Harmanec et al. [27]). Interactive RV measurements, based on the comparison of direct and flipped line profiles, were also carried out in reSPEFO2, independently by Petr Harmanec and Pavel Doležal to obtain some estimates of measuring errors.
We carried out a careful compilation of all available RV measurements of Tau with known dates of observations from the literature. These were complemented by our own RV measurements. In cases when the original sources give the dates and times of mid-exposures we used the programme HEC19 to derive heliocentric Julian dates (HJDs).5 The log of all RVs used in this study is in Table 1, while the log of all electronic spectra at our disposal is in Table 2. For convenience of future investigators, we publish all mean individual RVs, both from the literature and measured by us in (electronic only) Table A.1.
Some comments on individual data files follow. The files are identified by the spectrograph Nos. given in Table 1 and Table 2.
Spg. 1: Belopolsky [2] tabulated Pulkowa times and RVs in geographic miles. We converted his mid-exposure times to HJDs noting that G.M.T. = Pulkovo time - 2.0219 h. Using a Belopolsky’s paper on Lyr, where he tabulated RVs in both, the geographical miles and km s, we found that he adopted a conversion factor 1 geographic mile = 7.4211 km and used it to convert his RVs to km s. He reported two RVs for the secondary, but these are quite out of the secondary RV curve, and we omit them.
Spg. 2: von Aretin [55] published a very detailed analysis of 23 prismatic spectra from the Potsdam refractor. Since he tabulated his RV measurements for the comparison and stellar lines (carried out twice as measurements A and B, respectively, reversing the position of the plates on the comparator), we were able to recalculate the wavelength fits and derive RVs of stellar lines using modern wavelengths. When doing so, we noted the following obvious misprints in the tabulation: On page 106 in ‘measurement B’ the comparison lines 14 and 15 are actually 15 and 16. On page 112, measurement A, the stellar line 13 is actually line 11. On page 114, measurement A, the position of line 7 should be 22.667 (not 27.667 as printed). In principle, up to 14 stellar lines of H I, He I, and Mg II between 3889 and 4923 Å could be measured. However, depending on the plate quality, only some of them were actually measured on each particular plate. With two repeated measurements, this constitute up to 28 individual RVs for each plate. Therefore, we derived robust mean values for each plate and using the corresponding root-mean-square (rms) errors, we derived plate weight inversely proportional to (rms).
Spgs. 3&4: Schlesinger [45] published mid-exposures and RVs for 126 spectra obtained at the Allegheny Observatory. We used his mid-exposure times to derive the corresponding HJDs. Of these spectrograms, 98 were later re-measured by Ebbighausen & Struve [15], who used modern laboratory wavelengths. They also measured four earlier Allegheny spectra, not measured by Schlesinger, and 5 other, obtained after Schlesinger published his studies. The total number of spectra measured by Ebbighausen & Struve [15] is thus 107. There are 28 spectra, which were not re-measured by Ebbighausen & Struve [15] and for which we have to use the original Schlesinger RVs, based on older wavelengths. That is, why we assigned them a separate spg. number 3 to allow for a small difference in the velocity zero point with respect to spg. 4 with RVs based on Ebbighausen & Struve [15] RV measurements. We note, however, that there is no large difference between the original and re-measured RV values.
Spg. 5: McLaughlin [34] published RVs measured on the 1-prism spectra from Ann Arbor Observatory. We just derived HJDs from the published dates of observations and adopted his measured values.
Spg. 6: RVs measured on the prismatic spectra from the 1.83 m reflector of the Dominion Astrophysical Observatory were adopted from the study by Ebbighausen & Struve [15].
Spg. 7, 8, and 9: Ebbighausen & Struve [15] obtained 2 Lick spectra during two observing nights but published only mean RVs for each night. Similarly, they also published 68 nightly normal RVs based on RVs from 350 Mt. Wilson spectra. There is no way to distinguish Mt. Wilson and Lick spectra in their Table 5 so we assign the same zero point to them. In addition, we also append a single RV published by Zentelis [59].
Spg. 10: Fekel & Tomkin [16] used the McDonald 2.7-m coudé spectrograph and a Reticon detector with 1024 pixels. The covered spectral region was 105 Å centred on 4510 Å. The resolution was 0.45 Å = 30 km s per 1 pixel, and most of the spectra had a ratio of 400–500. Fekel & Tomkin [16] published individual He I 4471 Å and Mg II 4481 Å RVs for the primary, and Mg II 4481 Å RVs and mean RVs of several Fe II and Ti II lines for the secondary.
Spg. 11: We have at our disposal only the red segments of the CCD spectra from the CORALIE echelle spectrograph attached to the ESO LaSilla Euler Swiss 1.2 telescope.
Spg. 12: CCD spectra from the FEROS echelle spectrograph attached to the ESO LaSilla 2.2 m Cassegrain reflector (Kaufer & Pasquini [30]) were downloaded and preliminary reduced, rectified, and cleaned from cosmics by Jana Ščrková.
Spg. 13: HARPS CCD spectra from the echelle spectrograph attached to the ESO LaSilla 3.6 m reflector (Mayor et al. [33]) were also downloaded and preliminary reduced by Jana Ščrková.
Spg. 14: Red and a few blue spectra obtained with a coudé CCD spectrograph attached to the 2.0 m Ondřejov reflector were secured by various observers. They were calibrated in wavelength and converted to 1-D frames by Miroslav Šlechta in IRAF.
Spg. 15: Red CCD spectra obtained with the Littrow LHIRES III spectrograph attached to the Lisbon 0.356 m reflector were secured and preliminary reduced by Jose Ribeiro.
– Spg. 16: Two dense series of CHIRON echelle spectra ([54]), both covering the whole 33 d outer period, were secured by us at the CTIO in 2021 and 2022. Their preliminary reduction (wavelength calibration and creation of 1-D images) was carried out with the pipeline of the CTIO Observatory. Their further reduction, i.e. the order-by-order normalisation and creation of combined complete spectra with properly treated overlap between the blue orders, was carried out with the programme resimplex written by Pavel Doležal6 (see also Appendix A in Brož et al. [11]). The removal of cosmic spikes and occasional flaws and preliminary RV measurements were carried out with the programme reSPEFO2.7
– Spg. 17: Skalnaté Pleso 1.3 m Nasmyth-Cassegrain reflector with an echelle spectrograph ([39]) was used. The preliminary reduction of the spectra was carried out by Theo Pribulla.
Photometry
Photoelectric observations used in this study consist of observations secured in non-standard systems, observations, and observations. Their overview is in Table 3, and in the following we provide details on individual datasets. There are also test observations from the MASCARA survey ([52]) and from a CCD camera from the Ali Observatory in Tibet ([58]). These later were kindly put at our disposal by Jia Yin. They show eclipses in proper phases, but their scatter is rather high, and we do not use them here. All individual observations are provided in electronic only Tables B.1 to B.8 deposited to CDS.
– File A: Baxendell [1] Accurate times of these discovery visual observations are not known. We only know that Baxendell observed brightness decreases of Tau on the nights 1848 Nov. 6, 10, 14, 18, and 22. He mentions that the brightness of Tau was comparable to that of Tau = HD 25490 while it was much brighter than it on 1848 Nov 7. Since Tau has , it is clear that Baxendell indeed observed the primary minima on the five nights mentioned above.
– File B and Table B.1: Stebbins [50] These are the first photoelectric observations of Tau obtained with a photoelectric cell designed by Jakob Kunz and attached to 0.305 m refractor of University of Illinois Observatory. Stebbins observed Tau alternatively relative to five different comparison stars: Tau = HD 21364 (which is now known to be a eclipsing binary), Ori = HD 31237 (which Stebbins himself identified as an ellipsoidal variable with a period), Ori = HD 30836, Tau = 30 Tau = HD 23793 (which is suspected of variability under name NSV 1327 but no obvious variations are seen in the Hipparcos photometry by Perryman & ESA [38]), and Tau = HD 26912. For observations relative to 30 Tau and Tau, Stebbins attenuated the light of Tau by attenuating filters. He, however, measured the amount of this attenuation as well as the mean magnitude differences between individual comparison stars. Moreover, he corrected his observations relative to Ori for its known periodic light variations. After all these corrections, he tabulated all magnitude differences relative to Tau versus HJDs. As a precaution, we treated observations of Tau relative to those five comparisons as separate datasets and allowed PHOEBE to derive individual zero points for them. We omitted the following observations: Observations relative to Tau: HJD 2421252.522 and .542 (near primary minima of Tau), HJD 2421257.977, and 2421534.799 (near secondary minima of Tau); one observation relative to Tau: HJD 2421529.778, and one observations relative to Ori: HJD 2421296.547 (poor sky). We note that the numbers of observations in Table 3 give the original number of observations as well as the actual number of observations, which we used.
– File C and Table B.2: Nikonov [37] These first photoelectric observations secured with photomultipliers were obtained with a Guthnick photometer at Abastumani in seasons 1935 and 1936, and on MtCanobili with a Russian photometer in 1937. No colour filters were used and the photomultipliers clearly have the maximum of sensitivity in the blue region. Tau (HD 26912) and Tau (HD 25490) served alternatively as comparisons. All 1935 observations were obtained relative to Tau. Because of (an unconfirmed) suspicion that Tau could be a CMa variable, Nikonov finally reduced observations of Tau relative to Tau. Since the mean measured magnitude difference Tau - Tau were in 1935 but in 1936 (no value was given for 1937), we treat the observations from the three seasons as separate datasets. We omitted the following deviating observations: 2428060.444, all three from 28069, 28100.466, all from the nights 28113, 28114, and 28115 (clouds), and 28437.545. There are also two obvious typographical errors; observation 2428100.466 on page 32 should have a magnitude difference 0.271, not 0.371, and on page 34 of the paper the correct HJD of the first observation should read as 2428815.506.
– File D and Table B.3: Grant [20] These observations were obtained with a photoelectric photometer attached to the McDonald 0.33 m reflector. These are the first systematic observations covering phases of both minima, which were secured in and transformed to the standard system by the author. Unfortunately, they are tabulated relative to HR 1194 = HD 24155, which is now known as a CP2 variable V766 Tau. There is an obvious typographical error on page 84 of the paper: The correct HJD should be 2435101.656 (not .756 as printed). Besides, all observations for the nights HJD 2435132 and 2435144 are printed with incorrect sign. However, the greatest puzzle is that if one adds the observed magnitude of HR 1194 to the magnitude difference Tau – HR 1194, Tau appears as a sixth magnitude star! Conceivably, Grant used some attenuating filter when observing Tau but no such piece of information can be found in the paper. As it follows from a note on the first page of his paper, Grant died before the paper was finished and somebody else was preparing it for publication. There is no clue even in his PhD Thesis. At the end, we simply assumed that Grant used a attenuating filter. Then his observations agree well with other datasets.
– File E and Table B.4: Perryman & ESA [38] Hipparcos photometry with error flags 0 and 1 was used.
– File F: STEREO photometry ([4]) represents a very rich but relatively noisy dataset. For that reason we only used it to confirm our linear ephemeris derived from the global solution for all other photometric data.
– File G and Table B.5: this paper Photoelectric photometry of the star have been collected with a photometer attached to the 0.65 m reflector of Hvar Observatory in Croatia.
Figure B.1. Phase plots using ephemeris (2) for data excluded from our analyses. Top: File F STEREO photometry. Bottom: File K photometry from Tibet.
The data were carefully reduced to the standard system via frequent observations of Johnson standards and non-linear transformations and are tabulated for HJDs.
File H and Table B.6: this paper A series of BVRI CCD observations of Tau relative to HR 1243 = HD 25330 was obtained by Robert Uhlář at his private observatory at Jílové u Prahy with a G2/KAF0402ME CCD photometer attached to a 0.075 m reflector. These observations were reduced and transformed to the standard Johnson BV and Cousins RI filters using the C-MUNIPACK software.8
File I and Table B.7: MOST 2016 and 2017. These observations were downloaded at the Vienna tracking station, reduced by Thomas Kallinger and tabulated versus HJDs.
File J and Table B.8: this paper TESS photometry from sectors 05, 32, 42, 43, 44, 70, and 71 was downloaded and converted to magnitudes, using an arbitrary zero point, close to Johnson V magnitude. Barycentric BJDs are tabulated for these observations.
File K: This all-sky RGB photometry was obtained as a test case during commissioning the Ali Observatory in Tibet by Yin et al. [58] and was kindly put at our disposal by the author. We did not use it directly in our global photometric solution but only to confirmation of our new linear ephemeris.
References
- [1]Baxendell, J. 1848, MNRAS, 9, 37
- [2]Belopolsky, A. 1898, Astronomische Nachrichten, 145, 281
- [3]Berdyugin, A., Piirola, V., Sakanai, T., Kagitani, M., & Yoneda, M. 2018, A&A, 611, A69
- [4]Bewsher, D., Brown, D. S., Eyles, C. J., et al. 2010, Sol. Phys., 264, 433
- [5]Bolton, C. T. & Grunhut, J. H. 2007, in IAU Symposium, Vol. 240, IAU Symposium, ed. W. I. Hartkopf, P. Harmanec, & E. F. Guinan, 66
- [6]Borkovits, T. 2022, Galaxies, 10, 9
- [7]Borkovits, T., Csizmadia, S., Forgács-Dajka, E., & Hegedüs, T. 2011, A&A, 528, A53
- [8]Borkovits, T., Rappaport, S., Hajdu, T., & Sztakovics, J. 2015, MNRAS, 448, 946
- [9]Borkovits, T., Rappaport, S. A., Mitnyan, T., et al. 2025, A&A, 695, A209
- [10]Breiter, S. & Vokrouhlický, D. 2015, MNRAS, 449, 1691
- [11]Brož, M., Doležal, P., Vokrouhlický, D., et al. 2026, A&A, 711, A297
- [12]Casini, C., Galeotti, P., & Guerrero, G. 1968, IBVS No. 257, 1
- [13]Çelik, E. & Kahraman Aliçavuş, F. 2026, arXiv e-prints, arXiv:2609.08004arxiv.org/abs/2609.08004
- [14]Cester, B., Fedel, B., Guiricin, G., Mardirossian, F., & Mezzetti, M. 1978, A&A, 62, 291
- [15]Ebbighausen, E. G. & Struve, O. 1956, ApJ, 124, 507
- [16]Fekel, Jr., F. C. & Tomkin, J. 1982, ApJ, 263, 289
- [17]Georgakarakos, N. 2003, MNRAS, 345, 340
- [18]Georgakarakos, N. 2009, MNRAS, 392, 1253
- [19]Giménez, A. & Bastero, M. 1995, Ap&SS, 226, 99
- [20]Grant, G. 1959, ApJ, 129, 78
- [21]Hadrava, P. 1990, Contributions of the Astronomical Observatory Skalnaté Pleso, 20, 23
- [22]Hadrava, P. 1995, A&AS, 114, 393
- [23]Hadrava, P. 1997, A&AS, 122, 581
- [24]Hadrava, P. 2004a, Publ. Astron. Inst. Acad. Sci. Czech Rep., 92, 1
- [25]Hadrava, P. 2004b, Publ. Astron. Inst. Acad. Sci. Czech Rep., 92, 15
- [26]Harmanec, P. 1988, Bulletin of the Astronomical Institutes of Czechoslovakia, 39, 329
- [27]Harmanec, P., Yang, S., Šlechta, M., et al. 2026, A&A, 709, A167
- [28]Horn, J., Kubát, J., Harmanec, P., et al. 1996, A&A, 309, 521
- [29]Johnson, H. L., Iriarte, B., Mitchell, R. I., & Wisniewski, W. Z. 1966, Communications of the Lunar and Planetary Laboratory, 4, 99
- [30]Kaufer, A. & Pasquini, L. 1998, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 3355, Optical Astronomical Instrumentation, ed. S. D’Odorico, 844–854
- [31]Kostov, V. B., Rappaport, S. A., Borkovits, T., et al. 2024, ApJ, 974, 25
- [32]Krpata, J. 2008, http://astro.troja.mff.cuni.cz/ftp/hec/SPEFO/
- [33]Mayor, M., Pepe, F., Queloz, D., et al. 2003, The Messenger, 114, 20
- [34]McLaughlin, D. B. 1937, Publications of Michigan Observatory, 6, 3
- [35]Nemravová, I. A., Harmanec, P., Bencheikh, J., et al. 2013, Central European Astrophysical Bulletin, 37, 207
- [36]Nemravová, J. A., Harmanec, P., Brož, M., et al. 2016, A&A, 594, A55
- [37]Nikonov, V. B. 1938, Bull. Astronomical Astrophysics. Obs., 2, 29
- [38]Perryman, M. A. C. & ESA. 1997, The HIPPARCOS and TYCHO catalogues (The Hipparcos and Tycho catalogues. Astrometric and photometric star catalogues derived from the ESA Hipparcos Space Astrometry Mission, Publisher: Noordwijk, Netherlands: ESA Publications Division, 1997, Series: ESA SP Series 1200)
- [39]Pribulla, T., Vaňko, M., Komžík, R., & Sivanič, P. 2024, Contributions of the Astronomical Observatory Skalnaté Pleso, 54, 43
- [42]Rappaport, S. A., Borkovits, T., Torres, G., et al. 2026, A&A, 714, A523
- [43]Renson, P. & Manfroid, J. 1979, Information Bulletin on Variable Stars, 1658, 1
- [45]Schlesinger, F. 1916b, Publ. Allegheny Obs., 3, 23
- [46]Schmidt, J. F. J. 1858, Astronomische Nachrichten, 47, 373
- [47]Sepinsky, J. F., Willems, B., Kalogera, V., & Rasio, F. A. 2007, ApJ, 667, 1170
- [48]Škoda, P. 1996, in ASP Conf. Ser. 101: Astronomical Data Analysis Software and Systems V, 187–189
- [49]Soderhjelm, S. 1975, A&A, 42, 229
- [50]Stebbins, J. 1920a, ApJ, 51, 193
- [51]Stebbins, J. 1920b, ApJ, 51, 218
- [52]Talens, G. J. J., Deul, E. R., Stuik, R., et al. 2018, A&A, 619, A154
- [53]Tody, D. 1993, in Astronomical Society of the Pacific Conference Series, Vol. 52, Astronomical Data Analysis Software and Systems II, ed. R. J. Hanisch, R. J. V. Brissenden, & J. Barnes, 173
- [54]Tokovinin, A., Fischer, D. A., Bonati, M., et al. 2013, PASP, 125, 1336
- [55]von Aretin, E. F. 1913, Astronomische Mitteilungen der Universitäts-Sternwarte zu Goettingen, 15, 1
- [56]Škoda, P., Hadrava, P., & Fuchs, J. 2012, in IAU Symposium, Vol. 282, From Interacting Binaries to Exoplanets: Essential Modeling Tools, ed. M. T. Richards & I. Hubeny, 403–404
- [57]Wraight, K. T., Fossati, L., Netopil, M., et al. 2012, MNRAS, 420, 757
- [58]Yin, J., Yao, Y., Qian, X., et al. 2025, MNRAS, 537, 617
- [59]Zentelis, N. 1983, A&AS, 53, 445






