1 Introduction

The evolution of asymptotic giant branch (AGB) stars towards the planetary nebula (PN) phase of evolution is characterized by a series of helium shell flashes accompanied by enhanced mass-loss (Vassiliadis & Wood 1994). Thus the AGB stars have initial main-sequence masses of 1 $M_{\odot }$ to 8 $M_{\odot }$, but they end up as white dwarfs with a mass of only $\sim$ 0.6 to 1.4 $M_{\odot }$. In the Vassiliadis and Wood models, much of the mass loss occurs in one or more superwind phases which take place towards the end of the quiescent phases of helium shell flash cycles. Therefore, we might expect that the mass lost and subsequently ionized during the PN phase would appear as a series of shells around the main inner body of the PN.

Such outer structures should be structurally modified in a rather profound way during the PN evolution itself. As a matter of fact, it is commonly accepted that planetary nebula shaping results from the interaction between the stellar envelope ejected during the AGB phase (slow wind), and the fast stellar wind driven out from the central star by radiation pressure during the PN phase itself. In this two-wind model (Kwok et al. 1978), Rayleigh-Taylor instabilities play a role in shaping the outskirts and even the inner regions as well (e.g. [Dgani & Soker 1997]Dg97; Dwarkadas & Balick 1998). Furthermore, bipolar outflows may be set up either as a result of the binary nature of the central star, or as a result of the rotation of the precursor star, assumed to be single. Now, the observations (Hua 1997 and references therein) obviously display a large variety of nebular shapes: from symmetrical homogeneous to irregular and knotty configurations, which could reflect either different stages of evolution (Kwok 1982) or the results of interaction of several mechanisms (Bond & Livio 1990; Pascoli 1992; Chevalier 1994).

Observationally, the presence of secondary structures around some well-known PNe was already suspected long ago (Duncan 1937; Millikan 1974; Kaler 1974). Nonetheless, with photographic plates, their detection remained a difficult task due to their extreme faintness. The advent of high-sensitivity CCD detector technology has made the search for such structures much easier. Observations based on the first generation of CCD detectors ( Jewitt et al. 1986; Chu et al. 1987) reported such detections and related the existence of multiple-shell to the evolution of planetary nebulae and their parent central stars. However, in general, these observations were obtained using filters of rather wide bandpasses, so that, for instance, the two [NII] emission lines contaminate the proper H$\alpha$ emission.

In this paper, we present a new set of deep CCD observations performed using very narrow-band interference filters to provide optimum sensitivity for the detection of such outer structures, if and where they exist. We have observed 22 galactic PNe, among them some type I PNe (Peimbert & Peimbert 1983) which, on the basis of their high helium and nitrogen contents, are thought to be derived from massive precursor stars. In their simulation of ellipsoidal PNe taking into account the pole-to-equator density gradient, Zhang & Kwok (1998) found that bipolar and even "butterfly'' PNe are likely to have massive progenitors. In the case of the Magellanic Cloud PNe, this has been observationally established to be true (Dopita et al. 1997). In addition to the detection of peripheral (and sometimes outstanding) structures lying far beyond the previously known boundaries of the PN, we also give absolute fluxes in emission lines (namely H$\alpha$, [NII]6583 Å and [OIII]5007 Å).