The Recurrent Nova V745 Sco had a Classical Nova Eruption in 1897 July

V745 Scorpii (V745 Sco) is a recurrent nova (RN) with three known eruptions, in 1937, 1989 (Schaefer 2010) and 2014. The eruption light curves all have a very fast rise, a peak at B = 10.5, a fade by 3 mag in 9 days, and a plateau in the light curve for a class of P(9). Spectroscopically, the nova is a He/N class with emission lines including ionizations up to [Fe xi] and FWHM 3600 km s⁻¹. In quiescence, the star varies between B magnitudes of 19.9 and 21.1, while the red end of the spectrum is dominated by an M8 III red giant.

For RN demographics, we need lists of eruptions for each RN covering many decades, hopefully with no lost eruptions. To this end, I have been extensively searching for lost eruptions of all 10 known Galactic RNe, having now discovered 9 old RN nova events (Schaefer 2010, 2023). For V745 Sco, I had previous made exhaustive searches from the Maria Mitchell plates, the Sonneberg plates, and a non-exhaustive search through the Harvard plates. Now with the recent completion of the DASCH program ¹ (Grindlay et al. 2011), I have access to all the Harvard plates that show the V745 Sco position, so I have now exhaustively searched for long-lost eruptions. For this search, I have gone to Harvard and examined many plates by eye, in the traditional search method.

The 1897 eruption of V745 Sco appears on plate B19719, see Figure 1. DASCH measured that the nova image has a J2000 position that is 06 away from the astrometric position of V745 Sco. This is with the radius of the image being 64. DASCH reports that the image has a small ellipticity of 0.27, where the long axis runs WNW to ESE. This ellipticity is cause by ordinary trailing of 15, and is shared by all neighboring stars. The point-spread-function is a smooth ellipse with a sharp edge and a saturated interior, just like all the surrounding stars.

Zoom In Zoom Out Reset image size

The plate was taken with the 8 inch Bache refractor at Arequipa in Peru. The plate is a 10 minutes exposure on 1897 July 2 (mid-exposure at HJD 2414108.6593). The blue-sensitive emulsion has a spectral sensitivity effectively identical to the modern Johnson B system, so that with the use of B magnitudes for nearby comparison stars from APASS, ² the resultant nova magnitude is in the modern B magnitude system. The five closest comparison stars are labeled in Figure 1. With these comparisons made by-eye from the original glass plates, I variously get a magnitude of either 12.5 or 12.6, so I measure B = 12.55 ± 0.10. DASCH has also derived a magnitude, with B = 12.61 ± 0.11. From the light curve template of Schaefer (2010), V745 Sco appears with B = 12.4 at a time 4 days after peak. So the 1897 peak is at HJD 2414104.

From my by-eye checks and from the DASCH lists, only three useful Harvard plates were taken during the time of the 1897 eruption. Plate A2606 shows an empty nova position to a limit of 17.2 mag. This plate was taken 32 days after the discovery plate, when the 2014 eruption light curve implies B = 18.85. Two plates, BO 841 and BO 842, both show the nova position empty to limiting magnitudes of 10.8, with these taken 1-night before the discovery plate and the same night as the discovery plate. Both non-detections are easily consistent with the known V745 Sco light curve peaking at B = 10.5 four days before the discovery plate.

The evidence for the 1897 eruption comes down to one image on one plate. So we have to be confident in rejecting all classes of artifacts and plate defects. For this, I have searched all asteroids that can get brighter than 14th mag and find that none was within 33 of the nova position at the time of the exposure. Further, any main-belt asteroid would have a retrograde trailing of 51 over the 10 minutes exposure, and such would produce an easily detectable extra-trailing that is not seen. A close look at the original glass plate by my highly experienced eye shows that the nova image is not any type of plate defect. For example, when the emulsion is side-illuminated from the back (so as to detect dents and scratches on the emulsion), the nova image looks just like all nearby stars. Examination of the wide area around the nova shows no images that are not cataloged stars, and this rules out many types of artifacts, including double exposures and ghost images. (Figure 1 shows a number of small black dots scattered around, with these being normal for the B plate series in that year, with these always being greatly smaller than the stellar point-spread-function, so there will never be any confusion from these artifacts.) And the nova is a highly significant image, so we can rule out any type of noise for grain clustering. These results constitute proof that the 1897 image is actually that of V745 Sco in eruption. Further, even stronger proofs come from the nova image having the exact position of V745 Sco and having the identical point-spread-function as nearby stars.

So V745 Sco has known classical nova eruptions in 1897, 1937, 1989, and 2014, with time intervals of 40, 52, and 25 yr. The average recurrence timescale is 39 yr, with a fractional rms of 35%.

The constancy of the recurrence time for V745 Sco can be compared to the constancy of its sister-RNe T CrB, RS Oph, and V3890 Sgr. T CrB has a recurrence time that is constant to 2% over the last millennium. This is in contrast to RS Oph, which has recurrence times varying from 9 to 26 yr for its 9 known eruptions, with a fractional rms of 39%. V3890 Sgr has three eruptions with intervals of 28 and 29 yr, for a fractional rms of 2%.

V745 Sco nova eruptions are fast and faint, so many eruptions can easily be lost. For example, any eruption occurring from November to February will be lost due to the solar gap, with the Sun passing through Scorpius each year. Schaefer (2010) estimates that the probability of detecting a V745 Sco eruption from 1890 to 1898 is 1.4%, and from 1899 to 1953 is 33%, and these entirely with the Harvard plates. Further, the notorious Menzel gap in the Harvard plates from 1954 to 1969 makes for a drop in the probability of eruption-discovery down to 1.4%. So, before the 1970s, we should consider ourselves lucky to have detected the two eruptions in 1897 and 1937. Indeed, with the best eruption-discovery efficiency of 33%, it is likely that one-or-two further eruptions from 1897 to 1989 were missed.

If V745 Sco has missed eruptions before 1989, they are most likely near 1917 and 1963. With undiscovered eruptions near 1917 and 1963, the mean recurrence time would be 23 yr, with a fractional rms of 13%.

DASCH has been partially supported by NSF grants AST-0407380, AST-0909073, and AST-1313370.