New high resolution SWP IUE spectra of NGC 40, NGC 6543, NGC 6826, and BD +30 3639 have been obtained during 1994 and 1995 under the proposal RA090. The observational data are reported in Table 1 (click here).

The original spectral resolution is about 0.10 Å. As discussed in detail in Paper I, the wavelength calibration error of the reduced spectra is of

km/s, while the flux calibration error is of about 4%

longward of

and of about 6% shortward of

. This applies to the high resolution SWP images taken after November 10, 1981 at the Goddard Space Flight Center and after March 11, 1982 at VILSPA. For the high resolution images before these dates, the calibration function of Cassatella et al. (1981) is used. In this case the errors are of 5% and 8% for wavelenghts longward and shortward of

, respectively.

**Table 1:** High resolution IUE spectra of central stars of PNe
Name	Sp. type	SWP	Date	Time	Exp	Variat.^*
				hh:mm	min	detected
NGC 40	WC8 (1)	51880	22 08 1994	18:30	135	YES
		52744	06 11 1994	12:50	328

NGC 6543	OfWR(H) (1)	51881	22 08 1994	23:16	80	YES
		55982	21 09 1995	20:20	115

NGC 6826	O3f(H) (1)	51871	21 08 1994	22:53	115	YES
		55981	21 09 1995	15:50	120

BD+30 3639	WC9 (1)	51870	21 08 1994	18:34	170	YES

Considering that the reproducibility limit of IUE is of 6% at

level (Bohlin et al. 1980), we accept that changes by more or equal 10% indicate true variations in the spectra to a

level.

To search for variations in the lines profiles and/or in the continuum, the following procedure has been used. The total number of the available spectra in each object, including those listed in Paper I, have been averaged. Each spectrum has then been rescaled to the average spectrum by a factor, which, if different from 1, reveals the existence of changes in the level of the continuum. These changes refer to the whole spectrum and can be called "photometric'' variations. By comparing the rescaled spectra in the regions of the relevant spectral lines, one can detect changes of the line profiles. When a variation, photometric or affecting a line profile, is larger than 10%, we conclude that the change is real.

To reduce the noise, preserving however the shape of the real spectral features, i.e. without degrading the actual spectral resolution, we did apply a smoothing procedure. This was possible taking advantage of the oversampling of about 5 times the spectral resolution, performed by IUESIPS (IUE Spectral Image Processing System).

After the filtering, some residual artifacts from cosmic rays and from the reseau marks remain. They have been identified and made evident with crosses in the figures, using the quality factor parameter. We thought this is safer than to remove them from the spectra with interpolation procedures.

The spectral intervals considered to compare line profiles were of

around the mean wavelength of the most conspicuous P Cygni lines known to occur in CSPN in the UV range:

NV 1238.82, 1242.80; OIV 1338.60, 1342.98, 1343.51; OV 1371.29; SiIV 1393.73, 1402.73; CIV 1548.20, 1550.77; NIV 1718.55 Å.

We have examined with the above procedure the new spectra reported in Table 1 (click here) and all the spectra listed in Paper I of the same four stars. Note that in Paper I we did not compare to each other "adjacent'' spectra, i.e. spectra taken within the same IUE shift, considering that their separation being of the same order of the lenght of the exposures, hardly changes could have been revealed between them. Anyhow not to lose this possibility, we have now included in the analysis all the available individual spectra of the four stars under study.

As with the minimum detectable variation in the edge velocity of the P Cygni lines, we judge that in the spectra of our stars it amounts to

, where the "edge velocity'' corresponds to the blue border of the observed profile.

In the last column of Table 1 (click here) we report where variations have been seen. This takes into account also the spectra considered in Paper I.

2. Observations and data analysis