Publications of the Astronomical Society of the Pacific 92:338-344, June 1980

MULTIFILTER PHOTOMETRY AND POLARIMETRY OF NOVA CYGNI 1978 (V1668 CYGNI)

W. BLITZSTEIN, D. H. BRADSTREET, B. J. HRIVNAK, A. B. HULL, AND R. H. KOCH Department of Astronomy and Astrophysics and Flower and Cook Observatory University of Pennsylvania, Philadelphia R. J. PFEIFFER Department of Physics, Trenton State College, Trenton Flower and Cook Observatory, University of Pennsylvania, Philadelphia AND A. P. GALATOLA Space Division, General Electric Company, Valley Forge Received 1979 November 12, revised 1980 March 3

The results from more than 2700 filtered photoelectric observations of Nova Cyg 1978, obtained at the Flower and Cook Observatory, are summarized. The nova's decline through about 5 magnitudes is documented and amalgamated with similar observations already published by other groups. Variability on time scales up to 0.08 day and with peak-to- peak amplitudes up to 0^13 were common. No short-term periodicity was found. In the two-color plane the variability of the color indices is highly nonthermal and perhaps shows an inflection of slope about the time that dust was reported from IR observations by another group. A distance of 3 kpc is suggested. A few linear polarization measures, taken before dust formation, are listed. From the large interstellar component, a net intrinsic polarization is derived for the post-dust stages. Key words: novae—photometry—polarimetry

I. The Photometric Observations umn lists the interval of observation, the second, third, Nova Cygni 1978 was observed on 20 nights with the and fourth columns give, respectively, the number of Pierce-Blitzstein simultaneous two-channel, pulse-count- measures, the nightly average, and the minimum stan- ing photometer mounted on the 38-cm refractor of the dard deviation of a single magnitude difference calcu- Flower and Cook Observatory. The comparison star was lated from the actual pulse counts on the assumption of BD +43° 4012 (SAO 51194). Since the nova and com- Poisson statistics, all for the red filter. The corresponding parison star are so close together on the sky, the correc- parameters for the green and blue observations appear in tions for the instantaneous differential extinction were the last six columns of the table. Figure 1 shows the tem- always small. Counting times were 0.0004 or 0.0005 day, poral variations of the differential light and color-index depending on the decay stage of the nova. The diameter curves. of each diaphragm was 60^, and great care was taken to Standardized photoelectric observations have been avoid field stars when foreground sky brightness was published by Baldinelli (1978), Bruch (1979), de Roux measured. (1978), Giuricin, et al. (1978), Lindgren (1979), Mallama The detectors are RCA 4509 photomultipliers (similar and Skillman (1979), Margrave {l91Sa,b,c,d,), and Mattei to commercially available RCA 8645 photocells), and TABLE I each is equipped with broad-band red, green, and blue filters cut from the same stocks. The natural system char- Natural System Characteristics of the Pierce-Blitzstein acteristics for the photometer are summarized in Table I for an unreddened AO V star. With the present detectors, Photometer for an Unreddened AO V Star neither channel of the photometer has been standardized to the UBV system. More than 2700 observations have been obtained, Filter λ er,-. 1 FWHM nearly equally divided among the three bandpasses. It is Red 651θΧ 86θΧ impossible to publish all these data but they exist as File IAU(27), RAS-60 in the RAS library (Breger 1979). Table Green 5330 780 II summarizes our available photometry: the first col- Blue 4330 720

TABLE II Summary of Differential Observations, (Nova Cyg 78 - ΒΓΗ-43Ο4012)

(J.D.(hel.)-2443000.) n_ Am η Am Δι a r r r g g % b 766.5968 - 766.6139* 8 -3,094 +0.022 8 -2.239 +0.018 -1.983 +0.011 767.5263 - 767.6598a 23 -3.016 .016 55 -2.067 .018 47 -1.969 .015 772.5332 - 772.7960 77 -2.393 .011 87 -1.251 .011 72 -1.234 .006 773.7174 - 713.1552* 12 -2.223 .028 14 -1.056 .029 14 -1.012 .023 776.5123 - 776.5668 25 -2.093 .011 27 -0.802 .011 27 -0.808 .007 777.6107 - 777.7835 75 -2.089 .008 75 -0.792 .007 75 -0.753 .004 778.5489 - 778.7739 81 -2.122- .008 79 -0.841 .007 75 -0.789 .004 780.5319 - 780.5835 21 -1.818 .008 21 -0.363 .007 19 -0.444 .004 791.5075 - 791.6574 72 -1.049 .007 68 +0.258 .008 82 +0.313 .005 797.4886 - 797.6662 76 -0.789 .010 82 +0.773 .013 94 +0.706 .009 799.6936 - 799.7226 10 -0.669 .012 11 +0.790 .015 9 +0.673 .010 802.5896 - 802.6491* 13 -0.516 .007 13 +0.906 .008 7 +1.045 .005 804.5131 - 804.6751 57 -0.454 .009 45 +1.133 .011 64 +0.994 .007 806.6775 - 806.7045 11 -0.366 .008 11 +1.159 .010 11 +1.008 .006 809.4942 - 809.6526 59 -0.344 .008 60 +1.075 .010 61 +0.933 .005 810.4860 - 810.6642 70 -0.327 .008 69 +1.163 .011 71 +1.028 .006 811.4834 - 811.6574 67 -0.257 .008 66 +1.245 .010 66 +1.094 .006 813.6517 - 813.6814 11 -0.169 .009 12 +1.401 .013 12 +1.252 .007 822.5035 - 822.6325 57 +0.142 .010 58 +1.789 .019 64 +1.522 .010 832.4696 - 832.6316 59 +0.258 + .010 60 +1.921 + .018 61 +1.652 + .008

* : variable transparency

(1978). These measures appear in Figure 1 and are re- It is not surprising that short-term variability, as inferred to the right-side ordinates. By personal judgment, dicated in the example of Figure 2, exists for several the Pennsylvania observations were placed in register nights. In order to test objectively for the incidence of with the standardized observations, and it is only in the variability, each filtered data set for each night was test- sense of this personally derived zero-point correction ed by the x2-statistic. The results of these calculations that the former observations can be referred to the right- are shown in Table III and may be summarized as fol- side ordinates. From previous and concurrent programs lows: 42 of the 60 data sets, designated by υ, show in- on main-sequence stars, it is known that the scales of the trinsic variability above the 99.95% confidence level; ten natural Pennsylvania and UBV systems differ by about data sets, designated by u:, are found to be variable in 1%. It is not known whether such a close correspondence the confidence interval 99.95% to 94%; eight short data exists for the emission-dominated radiation from a nova. sets, designated by dashes, are not shown to be variable In Figure 1 the open triangles represent photographic by the test. For all these calculations, the gauge variance estimates or observations to which neither zero point nor includes both the pulse-counting Poisson statistics and scale corrections have been applied. These are plotted in short-term atmospheric effects, which latter have been the figure only to indicate the timing of maximum light. estimated from the comparison-star measurements and As Mallama and Skillman have already noted, the light which increase the Poisson standard deviations by factors curve is that of a fast nova, and Pfau's (1976) t3 and 15- of 1.2, 1.2, and 2.8 for red, green, and blue, respectively. day calibrations yield MB ^ —8. No matter what the On two nights which are common to Pennsylvania and bolometric correction for an emission-dominated spec- Margrave data, the latter author also observed short- trum, this absolute magnitude exceeds the Eddington time-scale variability. For a third night in common. Mar- limit for a one solar-mass object. grave did not note short-term variability and the Penn-

TABLE III -3 - 6 X2-Test for Short-Term Variability -2 - 7 Am. V J.D.(hel)- Am Am Am, 2443000 ^ i ^ -1 - 8 766 ν: ν: 0- 9 767 ν ν ν + 1 - ■10 772 ν ν ν + 2 . jll -H-2- 773 ν: ν ν: A(b-g) ■ -.2 ■ .0 .0 ' ^ *88 ¿ ΕB-V 776 ν ν: ν: +.2 -+.4 777 ν ν ν Δί'Λο- -+ .8 778 ν ν ν V-R + 1.4 - - +1.2 780 ν: ν ν -+1.6 + 1.8- 791 ν ν ν

770 790 810 830 797 ν: ν ν JDlhel.l -2443000. FlG. 1—From top to bottom the panels show the green light curve and 799 - ν: - two-color indices as a function of time. The Pennsylvania observations are represented by the filled circles whose ordinates appear as the left- 802 ν ν ν side scales. The open circles represent photoelectric observations and the open triangles photographic estimates or observations from other 804 ν ν ν stations. The right-side ordinate scales are described in the text. 806 -

809 ν ν ν 810 ν ν: ν Am 811 ν ν ν 813 - ν 822 ν ν ν

832 ν ν ν

For any bolometric correction, the nova (if it is a 1 SA)(© object) remains above the Eddington limit on this night. For a given filter, the filtering duty cycle is of the order of 0.007 day. All data sets were inspected for period- .45 .50 .55 .60 .65 icity but none was found. For further purposes of this Fraction of JD text "time scale" is defined as the interval between suc- FlG. 2—The observations of JD2443832. The ordinate tick marks are spaced at 0"1!. The net standard deviation calculated from the theo- cessive maxima or minima of short-term light variability. retical Poisson statistics and from the differential atmospheric extinc- The histograms of time scales and peak-to-peak ampli- tion noise is smaller than the symbol diameter for the blue data and tudes associated with the short-term variability are equal in size to the svmbol diameter for the red and green observa- shown in Figure 3. Because of the small number of tions. The curves drawn among the observations are freehand ones. nights and the discontinuous nature of the observations sylvania observing interval is too short for a decisive we claim no significant dependence of the amplitude comparison. It may be noted that JD2443767 is the first and time-scale histograms upon bandpass and, for a giv- night for which short-term variability can be shown. By en bandpass, the amplitudes and time scales are not cor- this night the nova had declined by about G1?? in blue. related in any way. Observational selection and noise en-

"T I I I ι ι ι ι—ι—ι—I 1 Γ B-V .5 .6 -1-7 Amplitude • · ·· · N ··· · r ······ · »9······ · ·

···· Ni ······· ·· »······· ······ ···

N< ··

-I I L .00 .05 .10 (MAG.)

τ—I 1 1 1 Γ

+ + Cycle Length -3 -2 -1 .0 +1 +2 .3 .4 N. • · ··· · Δ(b-g) FlG. 4—The development in time of the two-color plane. The number code gives the interval in days after the time of maximum light (JD 2443764.1) assigned by Mallama and Skillman; point 39 is weak in ·· · the blue bandpass and is disregarded. The observations by other teams N! ·· · ·· ··· ··· are shown by open circles, with the exception of that of deRoux for ··· ··· ·· JD2443792.7, which is far off-scale to the top and is not plotted. A thermal trajectory in the sense of increasing temperature is indicated by the arrow. The locus drawn among the data points is a freehand Nr ·· · • ······ · ous remarks already published in I.A. (7. Circulars con- I I i i t J I L cerning spectral details and their changes are consistent .00 .05 .10 (DAYS) with the early color changes in that the emergence and strengthening of emission lines and bands will preferen- FlG. 3—The histograms of the amplitudes and time scales for the short-term variability among the data sets. The caret on the abscissa tially increase the fluxes measured with the red and blue for the cycle-length histogram indicates the length of the filtering duty bandpasses compared to that observed with the green cvcle. bandpass. It is also possible to calculate the development of a (B — R) index. Although this is not shown in Figure sure that small amplitude and very short time scales will 1, it may easily be inferred that, in the early decay have been missed; and, since the nightly observing inter- stages, the emission increase is greater in red than in val was never longer than about 0.18 day, the detection blue. The development in time of the correlation be- of longer time scales was impossible. Our conclusions for tween the color indices is shown in Figure 4. The scaling our green observations are supported by Margrave's of the standardized and natural system color indices is (1979) interpretation of his V data. Each data set was consistent with the paired ordinates of Figure 1. The also inspected for evidence of a secular trend through nonthermal character of the changes of the indices is the observing interval. There occur approximately equal clear from the difference between a typical thermal tra- numbers of increasing and decreasing trends in bright- jectory and the temporal development of the indices ness over the 66-day observing interval, and these trends themselves. Obviously, this behavior should be attributed are likely to be segments of the transition-phase os- to the emission features occurring within each bandpass. cillations of the nova decay. If there are brief episodes of purely thermal behavior in The two lower panels of Figure 1 show the devel- these color indices, they cannot be recognized from the opment of two broad-band color indices. The senses of present data. The first slope change of the development the color changes demonstrate that the nova generally in Figure 4 occurs near point 3 and may be associated decayed faster in green than in red or blue. The numer- with the Ha emission dominating the red bandpass.

Gehrz et al. (1978) commented upon the detection of Morrison (1971) have cautioned that novae emission dust in the nova envelope between JD2443789 (25 days lines could affect the efficiency of a depolarizer but after maximum light) and JD2443797 (point 33). The IR found no such effects larger than 0.05% for the stars fluxes by Phillips et al. (1979) begin to increase no later which they studied. As with Nova Cygni 1975 (Hull than this latter date. In the coding of Figure 4, this inter- 1977), we have found no signature of depolarizer failure. val is associated with the interval between points 22 and We have made no attempt to correct the instrumental 33. It is conceivable that the visible-band photometry bandpasses for the nova spectrum. The instrumental re- detects activities concurrent with the dust formation sponse cited in Table IV is a combination of the filter events, but point 27 shows that this is not an orderly transmissions, photocell sensitivities, telescope reflec- trend. A single linear fit is inadequate to represent the tions, and atmospheric attenuation, but the emission data points of Figure 4. spectrum of the nova will inflect these characteristics by Because short-term variability occurs for each band- unknown amounts. In terms of {pßE) parameters the pass, there is commonly short-term variability for each present observations are listed in Table IV, which gives Solor index. These variations were inspected, but no sig- the integration time in the final column. The ultraviolet nificant correlation among the indices endured over data point reported by Pfeiffer (1978) has internal incon- even one night. sistencies and is now withdrawn. Both nights of our mea- From measures of interstellar absorption features Ake, sures are characterized by high and variable extinction, Lanning, and Mochnacki (1978) and Slovak and Vogt and both have concurrent photometry cited earlier in (1979) have independently determined color excess, dis- this paper. tance, and absolute magnitude for the nova at maximum By an assortment of criteria—signal variability, depar- light. Figure 1 shows (β—V) ^ +0.65 at maximum. The ture of the polarization spectrum from the accepted results of Slovak and Vogt lead to {B—V)0 max = +0.27 range of interstellar spectra, deviation between the ob- and Mvmax = —7.8 at light maximum while the inter- served and interstellar angles—intrinsic polarization has pretation by Ake et al. gives (^-V)0max = +0.05, been asserted for Τ Pyxidis = Ν Pyxidis 1966, HR Del- Mj/max ~6·4. Since the 3-magnitude decline time is phini = Ν Delphini 1967, FH Serpentis = Ν Serpentis of the order of 26 days, Mv would be expected to be of 1970 by Zellner and Morrison, and V1500 Cygni = Ν the order of —7.8 by Pfau's (1976) criterion. This agrees Cygni 1975 by Hull and several other authors. For Ν with the value derived from Slovak and Vogt's distance. Cyg 1978, the case for an intrinsic signal is based on sev- Each value of {B — V)0 conforms to Pfau's (1976) inter- eral of these arguments. pretation of (ß — V)0 for numerous older novae. In addition to the Pennsylvania polarization data, we are aware of a single green or blue observation by Kraut- Π. Polarimetric Observations ter (1978) and eight nights of UBV measurements by The nova was observed polarimetrically on the nights Piirola and Korhonen (1979). The latter authors have es- of 1978 September 16 and 21 using the Pennsylvania tablished that the polarization changed in magnitude, two-channel polarimeter coupled to the 72-cm reflector. position angle, and spectrum over the interval of dust Both nights precede the time of dust formation reported formation. We have formed weighted means of all the by Gehrz et al. The instrument and procedure have been predust measures, and in Figure 5 compare these to the described by Hull (1977), and the small corrections for weighted means of Piirola and Korhonen's postdust re- the instrumental polarization were calibrated by both sults. If only the green-through-ultraviolet predust means null and nonnull polarization standards. Zellner and TABLE IV Pennsylvania Linear Polarimetry for Nova Cyg 78

(J.D.(hel.)-2443000.) X /FWHM eff Ρ

767.55 66I08/IOO08 1.7% + .1% 30 ±2 320 sec 767.58 5390 /850 1.5 .1 34 2 320 767.60 4330 /800 1.7 .1 30 2 320 772.694 3700 /320 1.9 .2 32 3 1920 772.792 3700 /320 1.8 .2 27 3 1920 772.794 3700 /320 1.4 + .2 33 +3 1920

eral times their mean errors, and the Pennsylvania values of θΕ differ systematically from the values observed at Tuorla. None of these measures is simultaneous with any other and a low-level, short-term variability is a possible interpretation of their behavior. Even if this is so, it is almost a certainty that—the red data excepted—the polarization measures before JD2443785 are dominated by the interstellar component. If the polarization in some bandpasses can be assumed to be interstellar, something more can be said. The measured values of ΘΕ and ρ can be compared to the interstellar gradients of these parameters, and even though there is some dispersion in the nova measures, a distance greater than 2 kpc is required. Furthermore if a lower limit of E{B—V) is established by E{B — V) > p(%)/9, then the photometrically derived distance may not ex- ceed 5 kpc. For convenience, a summary of the characteristics of the nova appears in Table V. Errors have been formally propagated, and it is seen that this object is very similar to other fast novae. We have profited from conversations with J. Patter- son, W. Hägen, and C. C. Wu and are indebted to the referee for valuable suggestions. Mrs. F. H. Jamieson prepared the typescript. This research has been supported in part by grant NSF 77-15247, which is grateful- Fig. 5—The solid symbols through which the solid curve (representing the interstellar spectrum described in the text) is drawn represent the ly acknowledged. polarization spectrum derived from the weighted Pennsylvania and REFERENCES Tuorla observations during the interval before dust formation. The open circles refer to the weighted Tuorla measures after dust forma- Ake, Τ. B., Laiming, Η., and Mochnacki, S. W. 1978, LA.U. Circ. No. tion. The open squares, among which a dashed curve is drawn, show 3272. the vectorial remainder and are presumed to represent the spectrum Baldinelli, L. 1978, l.A.U. Circ. No. 3278. of the intrinsic polarization of the nova after dust has formed. Breger, M. 1979, Pub. A.S.P. 91, 408. Bruch, A. 1979, Ccmmi. 27 LA.U. Inf. Bull. Var. Stars No. 1567. are considered, the data fit very well an idealized inter- Coyne, G. V., Tapia, S., and Vrba, F. J. 1979, A./. 84, 356. deRoux, J. K. 1978, Comm. 27 LA.U. Inf. Bull. Var. Stars No. 1519. stellar profile parameterized by ρ = 1.59% at Amax = Gehrz, R. D., Grasdalen, G. L., Hackwell, J. Α., and Ney, Ε. 0.49 μ. This value of Amax agrees well with the determi- P. 1978, LA.U. Circ. No. 3296. nation by Serkowski, Matthewson, and Ford (1975) for Giuricin, G., Mardirossian, F., Mezzetti, M., Pucillo, M., Santin, P., the nearby star HD 204710. A residual of 0.23% for the and Sedmak, G. 1978, LA.U. Circ. No. 3316. red measure suggests an early intrinsic polarization that Hull, A. B. 1977, Sov. Astron. 21, 368. Krautter, J. 1978, LA.U. Circ. No. 3270. presumably can be attributed to nonspherical scattering Lindgren, H. 1979, Comm. 27 LA.U. Inf. Bull. Var. Stars No. 1543. of Ha emission. Clearly, the postdust means cannot be Mallama, A. D., and Skillman, D. R. 1979, Pub. A.S.P. 91, 99. fitted to an interstellar spectrum. If the predust means Margrave Τ. Ε. 1978«, LA.U. Circ. No. 3281. are assumed to be interstellar in the UBV bandpasses, — 1978b, ibid. No. 3296. vectorial subtraction from the postdust means yields a 1978c, ibid. No. 3299. 1978c/, ibid. No. 3316. spectrum which is largely due to intrinsic mechanisms. 1979, Bull. A.A.S. 11, 390. Although no attempt has been made to model complex Mattei, J. 1978, LA.U. Circ. No. 3303. emission-line, geometrical, and optical depth factors, this Pfau, W. 1976, Astr. and Ap. 50, 113. intrinsic spectrum does not derive from electron scatter- Pfeiffer, R. J. 1978, LA.U. Circ. No. 3278. ing with hydrogen self-absorption. Phillips, J. P., Wade, R., Selby, M. J., and Sanchez Magro, C. 1979, M.N.R.A.S. 187, 45P. Although emphasis has been placed upon the polariza- Piirola, V., and Korhonen, T. 1979, Astr. and Ap. 79, 254. tion variability through the stage of dust formation, some Serkowski, K., Mathewson, D. S., and Ford, V. L. 1975, Ap. J. 196, equivocal evidence does exist to indicate intrinsic varia- 261. bility at a much earlier time. For a given filter, the range Slovak, Μ. H., and Vogt, S. S. 1979, Nature 277, 114. Zellner, Β., and Morrison, Ν. D. 1971, A.J. 76, 645. of Θε in the observations by Piirola and Korhonen is sev-

TABLE V Assumed and Derived Parameters for Nova Cygni 1978

Quantity Value Comments

max + 6.0 ±0.2 mag (B-V) + 0.65 ±0.05 mag (b) t 3 26 ± 3 days Based on Mallama & Skillman's (19 78) t max (V) = JD 244376A.1 B.max 3.1 ± 0. A mag Based on Pfau's (1986) calibration lyL· = -10.67 + 1.80 log t0(B) B'max ±.30 ±.20 3 (B-V) + 0.27 ±0.09 mag Based on Slovak & Vogt's (1979) spectroscopic determination 0 ,max E(B-V) = 0.38 ± 0.08

M,V, max 7.8 ± 0.4 mag 0.49 ± 0.02 μ Polarization spectrum consistent with Serkowski's ejt al. (1975) value for nearby star HD 204710

R(X1 max ) 2.91 ± 0.26 Coyne, Tapia & Vrba's (1979) calibration R = (5.94 ± 0.48) λ A(V) 1.1 ± 0.3 mag

V-Mv-A(V) 12.7 ± 0.5 mag

γ photom c'tric 3.5 ± 0.8 kpc rpolarimetric 2-5 kpc M (r jj · ) - 7.7 ± 0.4 mag Based on Slovak & Vogt's (1979) reddening estimate fcr V,max reddening r = 3.3 ± 0.6 kpc