OBSERVATIONS OF BINARY STARS WITH THE DIFFERENTIAL SPECKLE SURVEY INSTRUMENT. VI. MEASURES DURING 2014 AT THE DISCOVERY CHANNEL TELESCOPE

In our work at the WIYN telescope with DSSI, we constructed a slit mask for use in the determination of the pixel scale and orientation at that telescope. We were able to attach the mask to the tertiary mirror baffle support and, with the help of WIYN staff, were able to easily mount and remove the mask with minimal loss of observing time. This essentially follows the example of the CHARA and USNO speckle programs, and is generally considered the most reliable method for precise scale and orientation calibration for speckle measures. However, at the DCT, we do not yet have such a system in place since the tertiary mirror is located inside the instrument cube and therefore inaccessible. As a result, we were forced in 2014 to rely on a group of calibration binaries during each run. This limits the precision of our measures somewhat, especially at the larger separations that we report here, since the scale in arcseconds per pixel is multiplied by the separation in pixels to obtain the final separation. However, we will show that there is no evidence for a systematic offset in separations we obtain. In the future, we plan to develop a system by which a calibration mask can be placed inside the DSSI camera itself. This would allow us to make a robust scale determination from first principles regardless of the telescope with which DSSI is used, and will be important to ensure the long-term consistency of our speckle measures.

On each run, we selected a small number of binaries with extremely well-known orbits with which to measure the scale. We required that the the orbits used include recent data and that the ephemeris uncertainties in position angle and separation be less than or equal to 05 and 1 mas respectively. Table 1 shows the final scale in mas per pixel and the offset angle between celestial coordinates and the pixel axes that were applied on each run. Because we observed very few objects in November for this project, we did not have the opportunity to observe scale objects, so the results from the September/October run were applied.

Table 1.  Final Values for the Pixel Scale and Orientation

Run	Channel A	Channel A	Channel B	Channel B
	Offset Angle	Scale	Offset Angle	Scale
Mar	545	18.433 mas/pix	−577	20.486 mas/pix
Jun	527	18.249 mas/pix	−562	20.345 mas/pix
Sep/Oct	539	18.781 mas/pix	−539	20.029 mas/pix
Nov	539	18.781 mas/pix	−539	20.029 mas/pix

Download table as:  ASCIITypeset image

Table 2 shows the objects used to obtain the results in Table 1 for each run. The columns give (1) the month of the run, (2) the Washington Double Star (WDS) number of the binary used, (3) the discoverer's designation, (4) the Hipparcos number, (5) the Besselian year of our observation, (6–9) the observed minus calculated residual in position angle and separation for both the A and B cameras, and (10) the reference for the orbit used to do the calculation. Although we did not take data on scale calibration objects in November, we did however measure the scale in another way, albeit at lower precision. We took a sequence of 1 s exposures on a bright star, offsetting the telescope in various directions in between the exposures. By measuring the centroid position of the star in each case and calculating the shift between frames, we were able to compare this to the number of arcseconds we had moved the telescope, and thereby determine both the scale and offset angle. This method allowed us to conclude that the scale and offset angles in November were consistent with those derived for the September/October run. Therefore, we used the latter to all of the November data.

Two things may be noted from Table 2. First, for each run shown, the average residual is zero (or very near zero) for both position angle and separation in the A camera, and for the position angle in the case of the B camera. However, the average residual is not zero for the separation residuals for the B camera. Specifically, for the five measures used in March, the average residual is −0.26 mas, and for September/October, it is −0.56 mas. This is because in this channel, DSSI is known to have a scale distortion which is dependent on position angle that is related to the positioning of one of the optical components in the optical train. More information about this distortion can be found in Horch et al. (2009, 2011a). For the data presented here, we assumed that the offset of this element was the same as what we have measured at the WIYN telescope. If this parameter varied slightly from the WIYN value, it would be possible to obtain a non-zero residual for this purpose. By varying the scale, one can minimize, but not eliminate, an average residual in this case. This is not an issue for position angle or the scale in the A camera, which does not have this problem. Another way to check that the scale is not influenced by this is to compare results from the A and B camera for all observations; we show that these do not have a significant difference in Section 3.3.1. We conclude that the variations we see in Table 2 are likely dominated by random error.

Second, we may use the values in Table 2 to derive a simple estimate of the measurement precision by calculating the standard deviation of each type of residual in Columns 6–9, if we assume that the error in the ephemeris position is negligible. In this case we obtain σ_θ,A = 0.31 ± 006 and σ_θ,B = 0.26 ± 005 in position angle and σ_ρ,A = 1.7 ± 0.4 mas and σ_ρ,B = 2.5 ± 0.5 mas in separation. Averaging, this indicates internal precision of ∼03 in position angle and ∼2 mas in separation.

The reduction of science data followed the same sequence as described in previous papers in this series (e.g., Horch et al. 2011a, 2015). The autocorrelation function is formed for each frame of the speckle exposure, and these are summed. Near-axis subplanes of the image bispectrum are also calculated for each frame and summed. Then, using the method of Meng et al. (1990), we form an estimate of the phase function for the object in the Fourier plane. We combine this with the modulus of the power spectrum (derived from the autocorrelation function deconvolved by the observation of a point source) to have a full estimate of the Fourier transform of the object. This is low-pass filtered with a Gaussian function where the width is chosen to give an FWHM that is similar to that of a diffraction-limited point-spread function. Finally, to obtain a diffraction-limited reconstructed image, the result is inverse-transformed. To give an indication of the typical quality of the results, reconstructed images for two objects, HDS 2143 and CHR 74, are shown in Figure 2. Orbits for each object are presented later in the paper.

Zoom In Zoom Out Reset image size

The reconstructed image is then examined for companions visually. If a companion is noted, then the pixel location of the secondary peak is noted and used as the starting position for a fitting program that performs a weighted least-squares fit to the object's power spectrum. In the case of a successful fit, the measure has been added to our main list of results. If, on the other hand, a companion is not noted visually, we then compute a detection limit as a function of separation from the target. More about this process is given in the Section 3.4.

In Table 3, we present our measures of double stars. The columns are as follows: (1) the WDS number (Mason et al. 2001), which also gives the right ascension and declination for the object in 2000.0 coordinates; (2) the Aitken Double Star (ADS) Catalog number, or if none, the Bright Star Catalogue (i.e., Harvard Revised (HR)) number, or if none, the Henry Draper Catalog (HD) number, or if none the Durchmusterung (DM) number of the object; (3) the Discoverer Designation; (4) the Hipparcos Catalog number; (5) the Besselian date of the observation; (6) the position angle (θ) of the secondary star relative to the primary, with north through east defining the positive sense of θ; (7) the separation of the two stars (ρ), in arcseconds; (8) the magnitude difference (Δm) of the pair in the filter used; (9) the center wavelength of the filter; and (10) the FWHM of the filter transmission. The measures have not been precessed from the dates shown. One hundred fourteen objects in the table have no previous detection of the companion in the fourth Catalog of Interferometric Measures of Binary Stars (Hartkopf et al. 2001b); we propose discoverer designations of Lowell-Southern Connecticut (LSC) 2-115 here. (These numbers begin at 2 because LSC 1Aa1, Aa2 = HIP 103641 was discovered in previous work in collaboration with Lowell (Horch et al. 2012).)

3.3.1. Astrometric Accuracy and Precision

Since we have a camera that takes data in two colors simultaneously, we can use the independent results in each channel to make a determination regarding the intrinsic precision of our measures. In Figure 3, we show the residuals obtained when subtracting the position angle and separation obtained in Channel B (880 nm) from those obtained in Channel A (692 nm). Figure 3(a) shows the position angle result for all measures in Table 3; these show a negligible relative offset: the average difference obtained is 0.04 ± 0.11 degrees, and the standard deviation obtained from all measures is 2.24 ± 0.07 degrees. The latter is expected to increase as the separation decreases, since the linear measurement uncertainty orthogonal to the separation subtends a larger angle at smaller separations. If only measures with separations larger than 0.1 arcsec are included, then the average difference remains consistent with zero, namely −0.08 ± 0.12 degrees, and the standard deviation decreases to 0.51 ± 0.09 degrees.

Zoom In Zoom Out Reset image size

In Figures 3(b)–(d), the separation differences are shown as a function of average separation, average position angle, and average magnitude difference respectively. From the entire data set, we obtain an average difference between the channels of −0.39 ± 0.12 mas with a standard deviation of 2.57 ± 0.09 mas. Such an offset might arise from the distortion in Channel B of the instrument, which is known to depend on position angle. However, when plotting the residuals as a function of position angle (Figure 3(c)), we see no clear trend. This is good evidence that the distortion model applied is appropriate for DCT data; if this were not the case, then the residuals would have a sinusoidal trend with two complete periods covered in the full range of 360° in position angle. On the other hand, when plotting the data as a function of average separation or magnitude difference, we see that there is a slight negative trend to the residuals for the measures reported below the diffraction limit (Figure 3(b)) and for measures where the magnitude difference is large (Figure 3(d)). For measures above the diffraction limit and having average Δm < 3, we obtain an average A–B difference of = −0.07 ± 0.11 mas, with a standard deviation of 1.96 ± 0.08 mas. Overall, we conclude that the scale and offset angles applied to the data yield no significant offsets between the channels for the bulk of the measures presented, with the caveat that below the diffraction limit and at large magnitude difference, there may be small systematic offsets particularly in separation. It is also clear that at large magnitude difference, the internal precision of the measures degrades. The average difference in separation for measures below the diffraction limit is −1.20 ± 0.44 mas with standard deviation of 2.78 ± 0.31 mas, and for measures with average magnitude difference greater than 3.0 we have an average difference of −0.95 ± 0.34 mas with a standard deviation of 3.60 ± 0.24 mas. Thus, given the data at hand, the systematic offset in separation appears to be on the order of perhaps 1 mas in both cases. This effect has so far not been noticed at WIYN, and should be studied further when more data at the DCT is available.

Nonetheless, for the purposes of the present paper we have not corrected for this possible source of error and we will use the intrinsic precision for all measures derived from the first separation difference mentioned, 2.57 ± 0.09 mas. Since this is a difference formed from two independent measures of (assumed) equal precision, the standard deviation obtained will be $\sqrt{2}$ times the precision of an individual measure. So, we estimate that the intrinsic precision of separation measures is ${\sigma }_{\rho }=(2.57\pm 0.09)/\sqrt{(}2)=1.82\pm 0.06$ mas. This in turn suggests that the position angle precision should follow the relation ${\sigma }_{\theta }=\mathrm{arctan}({\sigma }_{\rho }/\rho ),$ which has value 02 at a separation of approximately 0.5 arcsec; this is reasonably consistent with Figure 3(a).

There are a number of binaries observed that have orbits of relatively high quality in the Sixth Catalog of Visual Orbits of Binary Stars (Hartkopf et al. 2001a) that were not used in the determination of the scale. We may use these to further judge the intrinsic accuracy and precision of the measures in Table 3. A listing of these objects is given in Table 4, together with the orbit information. We studied the position angle and separation values separately. To be used for the position angle study, we required that the orbit have published uncertainties in the orbital parameters, and that for the epoch of our observation, the propagated uncertainty in position angle was less than or equal to 4°. For the separation study, we again required published uncertainties in the orbital parameters, and that the propagated uncertainty in separation for the epoch of observation was less than or equal to 4 mas. Figure 4 shows the observed minus ephemeris residuals in both parameters plotted as a function of separation. In this case, to make the plots clearer, we have averaged the observed result in both channels prior to plotting. There are no noticeable offsets or trends in either coordinate. The average residual in position angle is 0.16 ± 0.21 degrees, whereas the average residual in separation is −0.90 ± 0.94 mas with standard deviation of 4.96 ± 0.66 mas. Many of the objects with the largest residuals are also those with smallest separations. The internal repeatability study suggests that the precision of these measures is probably not very different than those of larger separations. It is therefore possible that the uncertainties for the orbital parameters for these objects may be slightly underestimated. If, on the other hand, we consider only those objects with separations greater than 0.1 arcsec, then the average residual in separation is 0.47 ± 0.81 mas. Here, the standard deviation of residuals is 2.91 ± 0.57 mas, with the average ephemeris uncertainty being 1.96 mas. If we subtract the latter figure from the former in quadrature, we can obtain another estimate of the intrinsic measurement precision, where the result is 2.15 ± 0.80 mas, which is consistent with the value obtained from comparing the A and B channels.

Zoom In Zoom Out Reset image size

Table 4.  Ephemeris Positions and Residuals Used in the Astrometric Accuracy Study

WDS	Discoverer	HIP	Besselian	θeph	ρeph	${\rm{\Delta }}\bar{\theta }$	${\rm{\Delta }}\bar{\rho }$	Orbit Reference
	Designation		Year	(°)	('')	(°)	(mas)
00121 + 5337	BU 1026AB	981	2014.7581	319.1 ± 1.8	0.3474 ± 0.0088	−0.5	(−3.3)	Hartkopf et al. (1996)
00507 + 6415	MCA 2	3951	2014.7554	136.6 ± 9.8	0.0477 ± 0.0019	(−3.6)	−8.0	Mason et al. (1997)
01057 + 2128	YR 6Aa,Ab	5131	2014.7555	297.9 ± 2.2	0.1057 ± 0.0064	−0.3	(−1.4)	Horch et al. (2010)
01072 + 3839	A 1516AB	5249	2014.7554	24.9 ± 4.4	0.1479 ± 0.0024	(+4.6)	+5.4	Hartkopf et al. (2000)
01083 + 5455	WCK 1Aa,Ab	5336	2014.7581	42.3 ± 1.6	1.1127 ± 0.0475	−4.1	(−41.0)	Drummond et al. (1995)
02022 + 3643	A 1813AB	9500	2014.7582	130.2 ± 12.4	0.0393 ± 0.0036	(+2.3)	+0.7	Hartkopf et al. (2000)
02157 + 2503	COU 79	10535	2014.7582	38.6 ± 0.1	0.1933 ± 0.0014	+0.6	+2.0	Muterspaugh et al. (2010a)
02396 − 1152	FIN 312	12390	2014.7557	87.0 ± 0.3	0.0864 ± 0.0004	+0.4	−3.6	Docobo & Andrade (2013)
02424 + 2001	BLA 1 Aa,Ab	12640	2014.7556	95.4 ± 3.0	0.0540 ± 0.0013	−0.8	−0.8	Mason (1997)
04044 + 2406	MCA 13Aa,Ab	19009	2014.7583	123.4 ± 15.1	0.0377 ± 0.0034	(−8.3)	−8.5	Mason et al. (1997)
04357 + 1010	CHR 18Aa,Ab	21402	2014.7584	120.9 ± 1.4	0.1553 ± 0.0059	+0.2	(+0.1)	Lane et al. (2007)
04382 − 1418	KUI 18	21594	2014.7584	3.1 ± 1.6	1.1105 ± 0.0218	0.0	(−28.3)	Hartkopf et al. (1996)
06098 − 2246	RST 3442	29234	2014.7587	146.0 ± 3.3	0.1131 ± 0.0029	−1.5	+4.7	Hartkopf et al. (1996)
06154 − 0902	A 668	29705	2014.7587	341.8 ± 4.0	0.2439 ± 0.0075	−1.8	(+1.4)	Hartkopf et al. (1996)
06171 + 0957	FIN 331Aa,Ab	29850	2014.7586	197.2 ± 5.6	0.0565 ± 0.0014	(+2.3)	−1.2	Hartkopf et al. (1996)
08468 + 0625	SP 1AB	43109	2014.2183	196.0 ± 0.6	0.2706 ± 0.0018	−1.3	+2.3	Hartkopf et al. (1996)
13100 + 1732	STF 1728AB	64241	2014.2189	12.1 ± 0.02	0.2210 ± 0.0019	0.0	(−36.7)	Muterspaugh et al. (2010a) a
		64241	2014.4620	12.1 ± 0.02	0.1611 ± 0.0020	−0.2	(−43.7)	Muterspaugh et al. (2010a) a
13396 + 1045	BU 612AB	66640	2014.2189	243.7 ± 1.1	0.2245 ± 0.0026	−1.1	+0.7	Mason et al. (1999)
		66640	2014.4621	245.6 ± 1.1	0.2171 ± 0.0026	−0.9	+0.7	Mason et al. (1999)
17217 + 3958	MCA 47	84949	2014.2190	90.7 ± 1.4	0.0213 ± 0.0005	+12.3	+5.9	Muterspaugh et al. (2008)
		84949	2014.4623	125.4 ± 0.3	0.0474 ± 0.0006	+2.7	+3.3	Muterspaugh et al. (2008)
17247 + 3802	HSL 1Aa,Ab	85209	2014.2190	244.2 ± 2.2	0.0255 ± 0.0005	+3.3	−9.1	Horch et al. (2015)
		85209	2014.4623	38.7 ± 4.0	0.0120 ± 0.0003	+7.3	+10.2	Horch et al. (2015)
19091 + 3436	CHR 84Aa,Ab	94076	2014.7549	295.0 ± 1.9	0.0553 ± 0.0016	−1.8	+4.4	Farrington et al. (2014)
20329 + 4154	BLA 8	101382	2014.7551	161.4 ± 0.1	0.0155 ± 0.0001	−7.5	−4.6	Torres et al. (2002)
21109 + 2925	BAG 29	104565	2014.7552	146.4 ± 1.3	0.2454 ± 0.0009	−1.0	+3.4	Balega et al. (2010)
21135 + 1559	HU 767	104771	2014.7551	0.3 ± 2.1	0.1098 ± 0.0025	+2.6	−0.5	Hartkopf et al. (1996)
21501 + 1717	COU 14	107788	2014.7552	65.3 ± 0.2	0.2345 ± 0.0014	−0.2	−1.6	Muterspaugh et al. (2010a)
22388 + 4419	HO 295AB	111805	2014.7636	334.1 ± 0.2	0.3093 ± 0.0028	+0.4	−2.3	Horch et al. (2015)
22535 − 1137	MCA 73	113031	2014.7552	114.7 ± 6.1	0.0781 ± 0.0040	(−6.4)	−2.7	Mason et al. (2010)
23052 − 0742	A 417AB	113996	2014.7552	91.7 ± 0.9	0.1953 ± 0.0016	+0.7	−3.5	Hartkopf et al. (1996)
23411 + 4613	MLR 4	116849	2014.7608	140.5 ± 1.8	0.1262 ± 0.0013	−1.7	−2.6	Hartkopf et al. (1996)
		116849	2014.7636	140.5 ± 1.8	0.1262 ± 0.0013	−1.2	−2.4	Hartkopf et al. (1996)
23529 − 0309	FIN 359	117761	2014.7609	181.1 ± 5.1	0.0659 ± 0.0017	(+5.0)	−8.8	Mason et al. (2010)
		117761	2014.7637	181.0 ± 5.2	0.0659 ± 0.0017	(+5.0)	−8.9	Mason et al. (2010)

Note.

^aAlthough the separation uncertainty meets the criteria discussed in the text, this object will probably need an orbit refinement after passing through periastron, as discussed in Horch et al. (2015), so we have not used the separations here. This does not affect the position angles, as it is an edge-on system.

Download table as:  ASCIITypeset image

3.3.2. Photometric Precision

We investigate the photometric precision of our measures by first defining a parameter which should approximately scale with the ratio of the measured separation to the size of the isoplanatic angle, as we have in previous papers in this series. Let this parameter, which we call q', equal the seeing value for the observation multiplied by the measured separation of the pair. (Since the seeing should be roughly proportional to the inverse of the isoplanatic angle, the desired ratio is obtained, up to a scaling factor.)

In Figure 5(a) we plot the difference between the magnitude difference we obtain for each measure at 692 nm appearing in Table 3 and the value of the magnitude difference appearing in the Hipparcos Catalog (ESA 1997). This comparison is limited since the center wavelength of the Hipparcos H_p filter is significantly bluer than either filter we have used. Nonetheless, the closest comparison we can make is with our 692 nm data. The plot shows what we have seen in previous papers, namely that the residuals are flat out to a value of q' = 0.6 arcsec² and that, at larger values, the most reliable measures tend to have positive residuals. This is expected since, as the separation becomes larger, there will be a loss of correlated speckles between the primary and secondary speckle patterns, and therefore a systematically large value of the magnitude difference will be obtained. For this reason, if the measure has a q' value larger than 0.6 arcsec², then we have not included the magnitude difference in Table 3.

Zoom In Zoom Out Reset image size

Figure 5(b) shows the ΔH_p values with the lowest uncertainties plotted as a function of the Δm(692 nm) value we have obtained. We have eliminated objects which are giants, have indications in the Hipparcos Catalog of variability, and those with B − V > 0.6. The plot shows a basically linear trend of slope near 1, as expected. The scatter about the mean line is higher than, e.g., Horch et al. (2011a), but this is probably still due to the fact that we are comparing results in a blue filter with one in a red filter, whereas in the 2011 work, there were many measures at 562 nm, a much better match to H_p. There are a number of examples in Table 3 of objects that were observed twice; the average difference in the magnitude difference obtained for these objects is 0.10 ± 0.01 in the 692 nm filter and 0.13 ± 0.02 in the 880 nm filter. As these are subtractions of independent measures of the same intrinsic precision, we can estimate the internal repeatability of our magnitude differences of $(0.10+0.13)/2\sqrt{2}=0.08\pm 0.02$ mag, very similar to results we have obtained at the WIYN and Gemini North telescopes.

Similarly, there are two objects, HIP 75312 and HIP 75695, that we observed on three separate occasions. The average standard deviation of the magnitude difference in each filter for these two cases is 0.068 at 692 nm and 0.081 at 880 nm. Assuming this is not a significant difference, then we can average all values regardless of filter to obtain 0.075 ± 0.012, in excellent agreement with the derived value from objects observed twice.

There were 104 objects that were observed with both filters under good conditions and the images obtained were judged to be of high quality, but did not show any evidence of a companion. In these cases, we can estimate the limiting magnitude difference as a function of separation from the central star from the reconstructed images. Our method for doing so was described in Horch et al. (2011a), but briefly, we form annuli centered on the peak of the central star of width 0.1 arcsec and separated by 0.1 arcsec. In each annulus, we locate all local maxima and local minima. We compute the average value and standard deviation of the local maxima. We also compute the standard deviation of the absoluted value of the local minima. Typically, this value is similar to that of the local maxima, so we average these to obtain a more robust estimate of the standard deviation of all extrema, and then consider the 5σ detection limit to be the average value plus five times this "average" standard deviation.

Figure 6 shows the result of this calculation for two objects: one where no companion was detected, HIP 77986, which is the Be star 4 Her, and one where a companion was detected, namely HIP 113690 = LSC 104. (The first measurement of this system appears in Table 3.) In these plots, a detection-limit curve is traced out using twelve concentric annuli, from 0.1 to 1.2 arcsec. The curve plotted is a cubic spline interpolation, and it assumes that the curve has limiting Δm of zero at the diffraction limit. In the case of LSC 104, we see a point representing a local maximum at separation 0.13 arcsec that is below the detection limit curve, indicating that this peak has greater than 5σ significance. In contrast, for HIP 77896, no peaks appear below the curve, indicating that no second source was detected.

Zoom In Zoom Out Reset image size

It is interesting to compare the shape of these detection limit curves with Figure 1(a). The envelope of the points in Figure 1(a) has a similar form, which indicates that, for high-quality observations, DSSI observations generally will not miss secondaries that fall within the area defined by the typical detection limit curves, such as those seen in Figures 6(c) and, (d). Table 5 gives our final list of high-quality non-detections, with the detection limit obtained at 0.2 arcsec. This value is just above the "knee" in the curve, which gives a good reference point for the rest of the curve.

This system has spectral type F2 from SIMBAD and parallax of π = 10.26 ± 2.92 mas from van Leeuwen (2007). A previous orbit was calculated by Cvetkovic (2008) with period of 109 years and semimajor axis of 0.344 arcsec. However, our new observation shows that the system has a much shorter period and smaller semimajor axis. However, our orbit yields roughly the same total mass, with large uncertainty. The system appears to have executed nearly a full orbit since the first speckle observations taken of it in 1985 by McAlister et al. (1987). More data will be needed to confirm this orbital trajectory, but if correct, the smaller period makes the system much more attractive for future study.

The composite spectrum of this pair is given in the SIMBAD database as a K0III, and we derive a magnitude difference of approximately 2.5, which is slightly higher at bluer wavelengths. This suggests that while the primary is evolved, the secondary is still near the main sequence, and so a study could be done along the lines of Davidson et al. (2009) to determine the age and individual masses of the two stars from the photometry through isochrone fitting. Also, the orbit we calculate has very high eccentricity, which helps explain why there were no successful observations of the pair in the late 1990s: periastron passage occurred in 1997, with predicted a separation of 6 mas.

A well-known speckle binary whose first interferometric observations date from the CHARA speckle program in the mid-1980's, this object went unobserved from 1996 until the present work. The new data suggest that nearly half an orbit has been completed since the first speckle observation. The primary is probably close to an A0V star, and with a magnitude difference of perhaps 2–2.5, we estimate that the secondary is an F2V star. This would suggest a mass sum in the range of 4 to 5 M_⊙, which is lower than what is implied by the value of P and a we derive, together with the Hipparcos parallax of 5.04 ± 0.39 mas. We expect that the next decade will provide an opportunity to improve upon the orbit presented here since the separation will be above 0.1 arcsec and the position angle is expected to cover about 30°. We encourage other observers to put this object back on their observing lists for the next few years.

A first orbit for this system was reported recently in Tokovinin et al. (2015), but those authors assumed a quadrant flip of one data point taken by Horch et al. (2008). That was not an unreasonable assumption based on the relatively small magnitude difference, but the points presented here give support to an "alternate" orbit where no quadrant flips are needed and the period is about half of that derived in the earlier calculation. Given the revised Hipparcos parallax of π = 13.55 ± 0.43 mas, the new orbit gives a mass sum of 3.5 ± 1.0 solar masses, which is somewhat high for a system of composite spectral type F7, but with substantial uncertainty at this stage. In contrast, the Tokovinin orbit gives a mass sum of about 1 solar mass, which would seem to be too low by roughly the same amount.

The new observations of this object presented in Table 3 indicate that the orbit is eccentric, and that we observed the system near periastron (and below the diffraction limit of the DCT). The orbital elements obtained confirm that picture, yielding a total mass of 3.4 ± 0.5 M_⊙. The composite spectral type is F5 and the system has a magnitude difference of about 1, so that we might suggest that the individual spectral types are F4 and F9. If so, then the total mass would be on the order of 2.5 M_⊙, which is about 2σ lower than the value implied by the current orbital elements and the parallax of π = 13.96 ± 0.55 mas. However, because there is only one point near periastron at the moment, the orbital elements will of course be modified with time.

The data at hand suggest that this F8 pair with small magnitude difference is viewed nearly edge-on. The mass implied by the spectral type is perhaps 2–2.5 M_⊙, if the system were to have solar metallicity, again lower than that implied by the orbital elements. The metallicity of the system is −0.31 according to the Geneva-Copenhagen Catalog (Nordström et al. 2007), which would tend to lower the mass at a given spectral type, thus making the discrepancy worse. Nonetheless, given the large uncertainty in the dynamical mass, it is no worse than a 2σ difference at this point. The mass ratio in the Geneva-Copenhagen catalog is 0.9, so this at least is consistent with the small magnitude difference obtained from speckle measures.