## Introduction to ER UMa stars

PG 0943+521 (later named as ER UMa) was originally discovered as an ultraviolet-excess object and subsequently confirmed to be a cataclysmic variable by Green et al. (1986). Until very recently the object had been little studied. It was only in 1992 when the dwarf nova nature of this object was first noticed by M. Iida (Hurst, Poyner 1992; Iida 1994 ). Intensive visual observations have revealed that the object varies between visual magnitude 12.3 and 15.2 (Iida 1994; Hurst, Poyner 1992). The most peculiar photometric behavior of this object is the presence of standstill'-like phase following initial declines of some outbursts (Iida 1994).

The object stays in this phase for ten days or so at 0.5 to 1.0 mag fainter than the outburst maximum. These characteristics closely resemble those of short standstills of Z Cam stars.

However, another intriguing feature is the presence of short (but sometimes as bright as the long ones) outbursts with a maximum decline rate of 0.7 mag/d (Iida 1994). These short outbursts resemble normal outbursts of SU UMa stars in the sense of its rapid decline rate, although no previously known SU UMa stars have shown to exhibit the Z Cam-type characteristics at the same time.

## Earlier observations

Spectroscopic determination of the orbital period was undertaken by Ringwald (1993), who classified the object to be a nova-like star like UX UMa, and obtained a candidate period of 0.1997 d from the radial velocity study. However the determination of the period was difficult because of very small semi-amplitude of the radial velocity variation, presumably caused by the low inclination configuration of the binary.

## The 1994 Jan. superoutburst

PG 0943+521 was caught in outburst by Poyner (private communication) using a 40-cm reflector on Jan. 1.056 UT at visual magnitude 12.5, The object faded to {\it V}=12.9 on Jan. 7, then entered a 'standstill'-like phase with a slightly slower decline.

The course of the outburst was followed by the author on nine nights at Ouda Station until Jan. 18 at V=13.6. The technical details will be given in Kato and Kunjaya (1995) .

The total duration of the outburst was thus longer than 18 days. The characteristics of the standstill'-like phase of the present outburst perfectly match those described by Iida (1994).

From the V-band light curve on Jan. 7, 1994, is shown.

Recurring but rather complicated features with a full amplitude of 0.16 mag and a recurrence period of 0.066 d is evident. To our knowledge on dwarf novae, only superhumps can explain such large amplitude modulation during an outburst, but the profile of the light curve is rather uncommon to superhumps: gradual rise to the 'hump' maxima, more steeper decline, and dip-like minima.

The observations on the subsequent four nights confirmed this finding, which also shows double humps of unequal amplitudes.

The mean decline rate between Jan. 7 and Jan. 11 was 0.043 mag/d, which is three times smaller than those observed in superoutbursts of usual SU UMa-type stars. Such small decline rate seems to simulate the `standstill'-like activity of this object.

A period analysis using the phase dispersion minimization (PDM) method (Stellingwerf 1978) after heliocentric correction and removing the trends of steady exponential decline, yielded a period of 0.06549 +/- 0.00003 d.

The situation changed since Jan. 15 observation. The light curve showed single-peaked humps with a steeper rise and gradual decline, which are very characteristics of superhumps.

The amplitude was relatively constant, i.e. 0.14 mag. The mean decline rate increased to 0.081 mag/d. The characteristics of the latter half of this outburst resembles more closely typical superoutbursts of SU UMa-type stars. A period analysis using the PDM method yielded a period of 0.06573 +/- 0.00005 d. Remarkable change in the profile of the light curve makes it difficult to connect these two periods, so we leave them as inidividual values.

Although there might exist some controversy whether this cyclic variation really reflects superhumps, we have identified it as such because of the following reasons: 1) The amplitude of variation in the outburst stage is large (0.14 mag). 2) The period significantly varied (~0.4 %) during the course of the outburst. 3) No evidence of such large-amplitude variation was found during a short outburst.

To our knowledge, nothing other than superhumps can satisfy the above observations. We believe that a precise radial velocity study would prove this identification. This cataclysmic variable is thus classified as a new member of SU UMa-type dwarf novae with well-determined superhump periods.

## Interpretation

The most peculiar characteristics of this dwarf nova are found in the general light curve, and can be summarized as follows.

1) All the long-living outbursts (or "standstills") of this object are actually superoutbursts.

2) There are some short outbursts as bright as the superoutbursts.

3) The superoutbursts last about 20 days, and the length of the supercycle (time between two succesive superoutbursts) is 43 days. Together with this unprecedent shortness of the supercycle length, no previously known SU UMa-type dwarf nova has been shown to spend such a long time (about a half of a supercycle) in its superoutbursting stage.

4) The cycle length of normal (short) outbursts is about four days, which is the shortest established one among all the dwarf novae.

5) The full amplitude of variation is about 3.0 mag, which is exceptionally smaller than those of usual SU UMa-type dwarf novae.

All the above observations show that PG 0943+521 is an exceptional SU UMa-type dwarf nova, especially in its short supercycle length (~43 d) and long duration (~20 d) of a superoutburst. To explain this peculiarity, we have tried to estimate the mass-transfer rate of this system from the outburst amplitude.

The small outburst amplitude of this system (~3.0 mag), when compared with the typical amplitude (5.8 +/- 0.5 mag) derived from ten well-established SU UMa objects of similar orbital periods studied by us, implies that the quiescent accretion luminosity, which may also be roughly proportional to the mass-transfer rate, of PG 0943+521 is 7--20 times larger than usual SU UMa-type dwarf novae. Although this estimate should be verified by more detailed analysis of the contribution of the accretion disk, the hot spot, the white dwarf, and the secondary to the spectral energy distribution, we would like to suggest that the peculiarities of the light curves may be a consequence of the large mass-transfer rate. This suggestion was later confirmed by numerical simulation of the outburst light curve by Osaki (1994).

As is well known, the mass-transfer of CVs below the period gap is believed to be powered by the loss of angular mementum by gravitational wave radiation. Due to the limited range of white dwarf masses and Roche lobe-filling lower main-sequence secondaries, the expected effect of gravitational radiation strongly depends on the orbital period, and this is what is actually observed in the tight relation between log P and Mdot. To maintain a high mass-transfer rate in PG 0943+521, there should be other driving mechanisms of mass-transfer below the period gap, for example heating of the secondary after the nova explosion. Although this possibility is merely speculative, PG 0943+521 would certainly play a crucial role not only in understanding the secular CV evolution below the period gap, but also in testing the thermal-tidal disk instability theory as a bridging case between usual SU UMa-type systems and nova-like systems with permanent superhumps.