The IUE Spectrum of the Planetary Nebula NGC 6905

We present results from new IUE high-dispersion and archival low-dispersion spectra of NGC 6905. Ultraviolet emission lines of O VIII and possibly O VII are detected for the first time in NGC 6905. High-excitation lines of C IV, C V, O VI, [Mg V], [Ne III], [Ne IV], [Ne V], [Ar IV], and [Ar V] are also observed. Electron densities and terminal wind velocities are presented. Terminal wind velocities from O V and C IV lines are discussed. Variability (up to ≈25%) of the complex split He II λ1640 emission may be intrinsic or a function of position angle of the IUE camera entrance aperture. After reextracting the low-dispersion data, no significant changes between 1981 and 1991 were found for the strong O V λ1371, C IV λ1549, He II λ1640, and C III] λ1909 emissions. Many weak features not detected before have been identified in the archival low-resolution data. Some comparisons are made with UV spectra of Sand 3 and KPD 0005 + 6501. Preliminary results indicate O vm emissions in IUE spectra of NGC 2371–2, NGC 2867, NGC 5189, and NGC 5315, suggesting that these planetary nebulae, together with NGC 6905, RX λ2117 + 3412, and Sand 3, form a new “O VIII sequence.” The present work illustrates the need to reexamine older IUE spectra and demonstrates that today's improved data analysis is capable of improving greatly the results reported in earlier publications.


INTRODUCTION
The variable planetary nebula (PN) NGC 6905 (061.4[09.5 in the notation of et al. hereafter Acker 1992, has been studied extensively, but its ultraviolet charac-A92) teristics have received relatively little attention. The IUE low-dispersion spectrum of NGC 6905 was described by hereafter hereafter Johnson (1981, J81). Feibelman (1982, searched for variability in IUE spectra and showed F82) stratiÐcation e †ects for di †erent emissions from echelle lineby-line analysis (the new data for the peculiar asymmetric He II emission are discussed further in In a survey of°4.2.3). about a dozen O VI central stars, hereafter Bianchi (1989, derived the interstellar extinction E(B[V ) \ 0.15 B89) from the j2200 bump, as well as some other parameters.
The optical spectrum of NGC 6905 was studied in detail by Aller who also observed the object in the (1951,1968), red and near-infrared region & Swens- (Aller 1977). Swings son commented on the strength of the He II j4686 (1952) line, comparable to Hb. The broad Wolf-Rayet emissions were depicted by and show the narrow nebular Aller (1968) emissions superposed on broad stellar features. NGC 6905 belongs to the small group of very high excitation PNs that exhibit very strong emission lines of O VI near jj3811, 3834, the so-called O VI sequence deÐned by & Aller Smith (1969). Only about two dozen objects of this type are known, and some exhibit variability. Variability for NGC 6905 was demonstrated by hereafter Vorontsov-Velyaminov (1961, on the basis of a change in the stellar He II j4686/Hb VV61) intensity ratio from a value of 0.5 to 1.3 between the years 1945 and 1959. Optical spectra do not display P Cyg pro-Ðles, but the UV data show strong P Cyg proÐles for C IV and O V, while N V is very weak and its reality is doubtful.
Similarly, the N III] j1750 multiplet is weak, which suggests that NGC 6905 is nitrogen poor.
Radial velocities for several O VI PNs were measured Ðrst by Rubin, & Ford but di †er from that DÏOdorico, (1973) given by ([7.4 km s~1) for NGC 6905, based on radio A92 observations from et al. DÏOdorico et al. Schneider (1983). also measured the He II j4686 intensity as 102 on a scale of I(Hb) \ 100.
Expansion velocity Ðeld maps, derived from spectra taken at di †erent slit position angles of NGC 6905, were presented by & Hamzaoglu hereafter Sabbadin (1982, All emission lines show splitting, and the [O III] SH82). j5007 emission gives a peak-to-peak value of D100 km s~1, depending on position angle. A slightly smaller value was obtained for Ha. The systemic radial velocity given by SH82 is [10 km s~1. These velocities are important to the discussion of IUE data in°4.2.3. & Shaw hereafter listed several Kaler (1984, KS84) parameters for NGC 6905, including a K T eff \ 104,000 that is comparable with the other O VI planetary nebula nuclei (PNNs) of their study and is considerably higher than the 50,000 K given by or the 60,000 K derived by B89, J81 from IUE data. They also gave an intensity I \ 98 for the He II j4686 emission, compared to I \ 100 for Hb. NGC 6905 has the distinction of having the greatest optical O VI jj3811, 3834 equivalent width, 700 of any of the stars Ó, measured by KS84. et al. hereafter measured Stanghellini (1995, SKSD95) optical emission lines and located NGC 6905 on the H-R plane of log for comparison with a dozen other T eff , L /L _ "" O VI sequence ÏÏ PNNs and also commented on the strength of [Ne III] lines. Their value of interstellar extinction, c \ 0.93, is much higher than that inferred by B89 from the UV j2200 extinction bump.
A set of four monochromatic CCD images of the nebula and star was presented by and are reproduced Balick (1987) in superseding earlier photographic images by A92, Louise and & Hua as well as two CCD Ha (1982) Louise (1984), images by Jacoby, & Arendt The remarkably Chu, (1987). complex structure Ðrst reported by from visual Curtis (1918) sketches and photographs is conÐrmed by the modern images.
Nonradial g-mode pulsation periods of about 16 minutes for the central star were reported by Bond &Kawaler These (1990, 1993) Bond, Phillips (1989, hereafter mention that the central star of NGC 6905 PM89) is a binary whose separation is but this proved to be 3A .6, erroneous, and the nucleus is a single star (A. Mampaso, private communication).
& Apparao list NGC 6905 as a possible Tarafdar (1988) X-ray source for which they gave only an upper limit of detection, as observed with the Einstein satellite. The central star shares some of the UV characteristics observed in con-Ðrmed X-ray PNs, namely, P Cyg proÐles that imply strong stellar winds, and a C IV j1549/He II j1640 ratio greater than 1 (Feibelman 1994).
The present study is concerned with the bright main body of the nebula and its central star but ignores the outer, faint ansae. We demonstrate that NGC 6905 presents an extremely complex UV spectrum and identify very high excitation emission lines from new IUE high-dispersion spectra. and we determine electron densities and terminal wind velocity. The detection of O VIII emission features makes NGC 6905 the third member of a select group that may constitute an "" O VIII sequence ÏÏ of planetaries and suggests that NGC 6905 is an object worthy of further study.

OBSERVATIONS
The IUE archive contains 11 low-dispersion spectra and one (relatively short, 120 minute) short-wavelength prime (SWP) high-dispesion spectrum. Two deep SWP and the Ðrst long-wavelength prime (LWP) high-dispersion spectrum were obtained by the writer in 1995. The IUE Log of  Observations for NGC 6905 is summarized in  This  Table 1. shows the image sequence number in column (1) ; low or high dispersion in column (2) ; the date of observation in column (3) ; the exposure time in minutes in column (4) ; the space craft roll (SCR) angle in column (5) ; the position angle of the major axis of the large (10A ] 23A) entrance aperture in column (6) ; ad some comments in column (7).
The position angle (P.A.) refers to that of the major axis of the large entrance aperture and is determined from the space craft roll (SCR) angle. The P.A. \ 73¡ minus SCR angle. See NASA IUE Newsletter 47 (1992) for further details. All observations were taken through the large entrance aperture of the IUE cameras and were centered on the nucleus of the PN. The data were reduced by means of the interactive computer routines available to guest observers at the Goddard Data Analysis Center (IUEDAC) and made use of the most recent calibration procedures.

WAVELENGTH IDENTIFICATIONS
We rely on several sources for wavelength identiÐcations. These include Aller hereafter and(1951, 1968 Ó; resolved at j1980 in high-dispersion data. who identiÐed only about two dozen emission B89, features. Considerably more information can be extracted now from these early spectra by means of the improved data reduction techniques that have been developed during the past 18 years at the Goddard Data Analysis Center (IUEDAC). In the SWP region alone, more than 35 features are measurable, compared to only 12 lines, or blends of lines, listed by and In we show the newer J81 F82. Figure 1 archival SWP 41863 of 30 minutes exposure after it was reextracted as described in On this Ðgure, and all°4.1. subsequent Ðgures, the vertical scale shown is in units of ergs cm~2 s~1 A~1.
The LWR 10067 spectrum is shown in Figure 2. Although saturated, this spectrum is very useful to show weak features that are not detected on shorter exposures.
In we show the region jj2000È2400 of NGC Figure 3 6905, which is similar to that of the ultra-high excitation low-mass W-R star Sand 3, considered to be a We take this opportunity to advise caution in accepting data from the "" Ðnal archive ÏÏ at face value. Probably the vast majority of these data for stellar objects are trust-   worthy, yet caution must be exercised for spectra of extended objects and for exposures a †ected by cosmic radiation hits. Comparison of the three best unsaturated SWP spectra, 14401, 16703, and 41863, revealed an apparent dramatic change in the He II j1640 emission : a drop of almost 50% in the emission Ñux for SWP 41863 compared to the other spectra. After some time was spent trying to verify this change (which occurred for the j1640 line only), it became clear that the reprocessed SWP.mxlo Ðle used the automatic routine SWET (signal weighted extraction technique) that eliminates blemishes and cosmic radiation hits. However, this method may overcorrect and throw away good data in the process, as it did for SWP 41863. Furthermore, the spectrum was not as well centered in the aperture as the others. Therefore, it is mandatory to examine the line-byline Ðle and reextract the data by means of the routine BOXCAR to integrate all echelle lines that contain useful information. Thus, it was found that He II j1640 emission is of stellar and nebular origin and extends from echelle line 43 to 72. Because of a nearby cosmic-ray hit, much of the data was deleted by the routine SWET, resulting in a Ñux that was about half its true value. When the Ñux is measured for the full range of integration, its value is the same (within an error of^10%) as for SWP 13401 or SWP 16703. A more detailed description can be found in NASA 56 see also & Linsky IUE Newsletter (1996) ;Nichols (1996). In we present the observed wavelengths from Table 2A SWP 41863 in column (1) ; the laboratory rest wavelengths in column (2) ; the ion identiÐcation in column (3) ; the source of identiÐcation in column (4) ; the remeasured emission-line Ñuxes (not corrected for extinction) from the best four spectra in columns (5), (6), (7), and (8) ; and some comments in column (9). The strong lines are heavily saturated in SWP 13404, but this spectrum is useful for detection of the weaker lines.
contains similar information from LWR data. Table 2B LWR 10067 is used for the weak lines, although the stronger lines are heavily saturated in this spectrum. Many highexcitation emission lines of C IV, C V, O III, [Ne V], and [Mg V] are present in the UV spectrum of NGC 6905 and its central star. Footnote b in corresponds to lines Table 2 that were detected in high-dispersion data whose wave-length determinations are much superior to those from Ðnal archive low resolution, the reverse is true for Ñux measurements because Ðnal calibrations for high-dispersion spectra are not yet available.

T he High-Dispersion Data
Apparently no detailed discussion of the SWP 18366 spectrum has been published so far. The new SWP 55994, SWP 56067, and LWP 31209 spectra were obtained during 1995.
The LWP spectrum of NGC 6905 resembles that of Sand 3 in that the region jj2750È2810 is crowded with numerous emission features that are difficult to resolve (see F96a). Moreover, until recently Sand 3 was thought to be a unique object with ultraÈhigh excitation lines of O VIII (IP \ 871.4 eV) in the optical region but has been joined by objects like KPD 0005]5106, RX J2117]3412, and others (see Werner et al. and references therein). We report the Ðrst detec-1996, tion of the ultraviolet emission lines of O VIII near 1930 Ó and 2976 for NGC 6905 in Ó°4.2.5. Emission features that fall within a fraction of an angstrom of known transitions for the crowded region jj2750È2870 are identiÐed in but deÐnite identiÐ- Table 3, cation and Ñux measurements require higher signal-to-noise (S/N) data. Isolated measurable lines are listed in Table 4. We show a section of the LWP 31209 spectrum in Figure 4 to indicate the richness of [Ne V] lines and for comparison with the low-resolution A smoothed plot of the Figure 3. regions jj2755È2805 is shown in where some rest Figure 5, wavelengths of likely identiÐcations are indicated. This plot is shown primarily as guidance for future investigations, as the region is too crowded for IUE to resolve individual lines. Moreover, it is impossible to determine the level of the continuum for this wavelength region.

T erminal W ind V elocities
The signal-to-noise ratio (S/N) for the three co-added high-dispersion spectra is too low to determine a valid P Cyg proÐle for the C IV doublet, even though it is seen on all low-dispersion spectra. Therefore, we, rely on the mean of the three best (not saturated) low-resolution spectra for a determination of km s~1 as shown in  This is a section from 1500È1675 of the unsatu- Figure 6. Ó rated SWP 41863 ; the strong C IV and He II lines are shown truncated for better visibility of the P Cyg absorption of the C IV feature. Thus, the terminal wind velocity for NGC 6905 is near the upper limit encountered for PNs by & Patriarchi Perinotto hereafter (1991, PP91). Low-dispersion spectra also show a P Cyg proÐle for the O V j1371 line. The three co-added high dispersion spectra give km s~1 for the j1371 P Cyg The N V doublet does not exhibit a P Cyg proÐle in high dispersion, although interstellar j1238 and j1242 absorption components are present. The N V doublet is barely detectable in the low-resolution spectra, yet variablity cannot be ruled out.

Electron Densities
On Ðrst impression, the unsaturated SWP 18366 spectrum yields an approximate electron density, log N e \ 3.1 cm~3 from the C III] F(j1907)/F(j1909) ratio \ 1.44, based on the diagnostic curves of Feibelman, & Berring-Keenan, ton and the assumption that all Ñux is of nebular (1992) origin at T \ 10,000 K. However, the triple-peaked C III] proÐles shown in strongly suggest a contribution Figure 7 from the central star. Therefore, we deconvolved each C III] component into one stellar and two nebular Gaussian pro-Ðles and derived separate Ñuxes. This yields a stellar Ñux of 6.2 ] 10~13 ergs cm~2 s~1 plus a sum of the nebular contri-  and may also be present in (Feibelman 1995) NGC 5189 and NGC 2371-2 (in preparation), thus suggesting that such proÐles may be characteristic of O VIÈtype W-C objects.
The Si III] j1892 line is extremely weak in NGC 6905, so that the F(Si III])/F(C III]) diagnostic is unreliable for N e from low-dispersion data.
From LWP 31209 we determine log cm~3 N e    Kurucz 1991 ;Kelly 1979 ;Garcia 1965 ;Adelman 1977. consider this to be a more accurate value than log N e \ 5.4 0.2 cm~3 derived from the low-dispersion F(j1601)/ F(j2422]2425) ratio and diagnostics, NussbaumerÏs (1982) after the observed Ñuxes were corrected for E(B[V ) \ 0.15 adopted from The weakness of the [Ne IV] j1601 line B89. and blend with C IV j1595 adds to the uncertainty.
The N IV F(j1483)/F(j1486) diagnostic cannot be applied to NGC 6905 because of the weakness of the N IV] lines.
A mean density of cm~3 is given by N e \ 1100 Pottasch p. 300) for K. (1984, T e \ 12,000 In we summarize some of these parameters taken Table 5 from recent literature and from this work.

He II V ariability : Intrinsic or Position Angle ?
The He II j1640 emission poses an interesting question. Comparison of the three high-dispersion SWP spectra indicates a real change in the emission proÐle and intensity as shown in Optical variability of j4686/Hb was PN. The dual peak proÐle of the He II j1640 line could be interpreted as the emission consisting of a stellar and a nebular component, or two nebular components. The change in the three spectra may be intrinsic or may be due to di †erent angles of the major axis of the entrance aperture intercepting slightly di †erent regions of the nebula. We are not able to state unambiguously which one of these possibilities, or combinations of them, is responsible for the di †erence in line proÐle. The systemic radial velocity of NGC 6905 is given as [8.4^1.7 km s~1 by and as A92 [10 km s~1 by This value lies near the trough of the SH82. two components of the He II j1640.390 line for the three plots in in which the thin line represents SWP Figure 8, 18366, the thick one is for SWP 55994, and the medium one is for SWP 56067. Note that the deepest trough corresponds to neither the shortest (200 minutes) nor the longest (532 minutes) exposure. The position angles for these exposures were [24¡, [3¡, and [16¡, respectively, and the deepest valley between the peaks corresponds to the smallest P.A., [3¡. We interpret the two peaks as nebular emission rather than of stellar origin, as there is no peak near [10 km s~1, but we cannot rule out some stellar contribution. This is, however, in contrast to the optical He II j4686 line shown by Smith & Aller (1969), where a narrow nebular He II emission is superposed on a very broad stellar W-R emission. We should expect to see a strong stellar He II component, but this may be buried in the trough of the proÐle.
The nebular expansion velocity of 51.5 km s~1 is then determined from half the separation of the two peaks at 103^5 km s~1, in reasonable agreement with the expansion velocity of 47 km s~1 determined by from j5007. SH82 From the C III] j1907 and j1909 proÐles, we obtain a value of 55 km s~1.
gives a weighted expansion Weinberger ( Fig. 11) display a split proÐle similar to CPM93 that of the j1640 feature.
A multiplot of the strongest He II lines observed in LWP 31209, jj3203, 2733, 2511, is shown in velocity space in The proÐles for j2733 and j3203 are double Figure 9. peaked, as was noted earlier for j2734 by from low-F82 dispersion line-by-line analysis. However, when j3203 (or jj2511, 2733) is compared with the j1640.39 line from SWP 55994, they appear to be shifted because di †erent portions of the nebula are seen in SWP and LWP, as indicated by their respective P.A., [3¡ and ]63¡.

As
shows, the position angle (P.A.) of the major Table 1 axis of the entrance aperture varies strongly for the di †erent exposures. The range of P.A. for the SWP high-dispersion spectra di †ers by only 21¡ (from [3¡ to [24¡), but the LWP 31209 P.A. at ]63¡ is nearly orthogonal to the SWP spectra and may explain the di †erence of line proÐles seen in He II j1640 compared to those of He II jj2511, 2733, 3203.

Is there O VIII Emission Present ?
Very high excitation lines, such as O VIII (IP \ 871.4 eV) j6068 (n \ 10 ] 9) and j4340 (n \ 9 ] 8) have been con-Ðrmed in the very hot helium-rich white dwarf KPD 0005]5106 by Heber, & Fleming as well as Werner,  shown in that covers the Ó (M93), Figure 10 region of 1928È1934 The j1930 feature is also seen in the Ó. SWP range but the j1932 line is unobservable, as it falls into the echelle interorder gap mentioned above. shows the IUE region jj2968È2978. These fea- Figure 11 tures need to be reobserved at higher S/N than is presently possible with IUE.

T he O VII T riplet
The only O VII feature observable in the IUE wavelength range is the triplet jj1623. 63, 1638.30, 1639.87 (multiplet 8) given in Of the three transitions, only the j1638 emis-M93. sion is identiÐed tentatively in the co-added SWP spectrum shown in This feature was also observed by Figure 12.
F96a in IUE spectra of two other O VI PNs, NGC 2371-2 and NGC 5189.

V ariability in the UV Spectra
Except for the peculiar di †erence in the He II j1640 highdispersion proÐles that may be a function of position angle as described in emission lines from low-resolution°4.2.3, spectra show remarkably little variability : each of the strong emission lines of O V j1371, C IV j1549, He II j1640, and C III] j1909 shown in shows less than 5% Table 2 variation during the years 1981È1991. A possible exception may be indicated by SWP 16703, which was taken under high background radiation conditions. In this spectrum, almost all the weaker lines are stronger than corresponding lines in the other three spectra.
We assign little importance to the lines below j1200 and show the C IV j1168 emission in parentheses to indicate low sensitivity and large errors.
If we are to look for variability, it may be found in the much weaker lines, notably those of C IV and O VI, but the errors are quite large (^50%) for these Ñuxes. less, they may be indicative of PG 1159Ètype Ñuctuations or coronal winds that require further study.

4.4.
A New "" O VIII Sequence ÏÏ for PNs We conclude on a speculative note. The detection of O VIII emission lines in the IUE spectrum of NGC 6905 raises the question of whether there exists an "" O VIII sequence ÏÏ of PNs. At least two other CSPNs are deÐnitely known to exhibit O VIII lines at optical and UV wavelengths. They are Sand 3 and the very hot X-ray source RX J2117]3412. As might be suspected, all three are also deÐnite members of another small subset, namely, nonradial pulsating stars & Bond et al. (Ciardullo 1996 ;Bond 1993). Several other CSPNs are currently under investigation for O VIII emission in IUE spectra. Preliminary results indicate that NGC 2371-2, NGC 2867, NGC 5189, and NGC 5315 show O VIII emission. The Ðrst two deÐnitely are pulsators, but NGC 5315 is not thought to be a pulsator by Ciardullo & Bond (1996).
After this paper was submitted, the existence of the "" O VIII sequence ÏÏ was conÐrmed from UV and optical data for a total of 10 PNs hereafter (Feibelman 1996b, F96b).

SUMMARY AND RECOMMENDATIONS
Although NGC 6905 has been known to show variability in optical spectra, no variability was detected in O V, C IV, He II and C III] from low-resolution data between 1981 and 1991. In fact, these lines were found to be remarkably stable, to^5%, after the data were reextracted by the latest reduction routines. Variations in the He II j1640 line proÐle from high-resolution data may be intrinsic or a function of position angle of the major axis of the large entrance aperture.
Some variability may be present in weaker lines of C IV and O VI, possibly indicative of PG 1159Ètype activity, but these lines are an order of magnitude fainter than the strong emissions mentioned above and may have large errors (^50%). N V is very weak, if present at all. Long-term FIG. 12.ÈThe region of the putative O VII emission at j1638.30. This feature is present on the three individual spectra and this co-added plot. The very strong He II j1640 line is truncated in this plot, which also shows the O V j1643 emission. monitoring of the optical O VI doublet near 3820 and Ó other lines may be worthwhile to detect periodic spectral variations.
NGC 6905 displays one of the highest ratios of 2V exp versus for any PN when added to the plot of V edge Patriarchi & Perinotto (1991).
A terminal wind velocity of [3880^200 km s~1 for C IV j1549 was derived from the mean of the three best low-dispersion spectra, and a value of 2730^150 km s~1 for the O V j1371 line from three exposure-weighted coadded high-dispersion spectra.
Emission lines of O VIII at 1930O VIII at , 1932O VIII at , and 2977 were Ó detected in the IUE spectra of NGC 6905, and possibly O VII near 1623 and 1638 but these need to be conÐrmed Ó, by higher S/N spectra and by their optical counterparts.
Other high-excitation lines of C It should be worthwhile to look for optical O VIII j6068 emission and to search for X-ray emission from NGC 6905.
I thank the Goddard IUE sta † for obtaining the 1995 spectra under the Service Observing program, and the VILSPA team for the SWP 56067 spectrum obtained during the 19th year episode of IUE operations for program SA018.