The low resolution spectrum, shown in Fig. 1, shows typical features of a symbiotic star, viz. Balmer emissions with steep decrement, which are signatures of the hot component, and TiO bands, probably originating in the cool atmosphere of a red star. In addition to the features seen by Kilkenny (1997), we find Fe I, He I and Fe II emissions, and most notably the Ca II 8498, 8542, 8662 triplet in emission. Based on the wavelengths, and the absence of Paschen 14, we suspect that these emissions are truly Ca II and not the Paschen P13, P15 and 16 lines. The emissions at $\lambda$ 4730, 5197 and 6384 Å are tentatively identified as Mn I whereas $\lambda$ 5991 and 6300 Å probably correspond to O I. The narrow absorption lines of the cool-star, often seen in symbiotics, do not appear in our spectrum. The H $\alpha$ emission wings extend to velocities $\sim$ 1000 kms^-1. The spectral characteristics described above, especially the presence of the Ca II triplet in emission and the absence of He II emission, places V589Sgr as a special member in the class of symbiotic stars. The steep Balmer decrement, and the absence of He II4686 emission, implies a temperature for the hot component lower than in other members of the class.

**Table 3:** Equivalent widths and radial velocities of the main emission lines
Line	$W_{\lambda}$ (Å)	rv (kms^-1)
H $\alpha$	-143	13
H $\beta$	-40	40
H $\gamma$	-21	11
H $\delta$	-9	-46
H $\epsilon$	-8	-162
He I 5875	-8	14
He I 5015	-10	105
He I 4922	-9	60
Ca II 8662	-1
Ca II 8542	-4
Ca II 8498	-7
Mn I 4730	-5	4
Mn I 5197	-5	-2
Mn I 6384	-1	-3
Fe I 4384	-6	15
Fe I 4490	-6	8
Fe I 4602	-1	28
Fe I 5168	-6	17
Fe II 4628	-4	7
Fe II 5362	-2	18
Fe II 6433	-2	6
Fe II 6456	-2	8
Fe II 6516	-3	17
Fe I + Fe II 4232	-5
Fe I + Fe II 4414	-10
Fe I + Fe II 5316	-5
Fe I + Fe II 6417	-1
O I 5991	-2	-5
O I 6300	-3	19
unknown 4618	-2
unknown 4666	-3
unknown 5234	-4
unknown 5275	-3
unknown 5534	-2
unknown 6147	-2
unknown 6317	-2
unknown 7633	-10

Table 1 lists the equivalent width and the radial velocity for the main spectral lines. Heliocentric radial velocities were measured by fitting Gaussian functions to the profiles whereas the equivalent width was obtained by integration of the flux between the points where the profile merges with the continuum. Since the spectrum shows strong gradients and severe line blending, it is not an elementary task to determine the continuum, so the errors associated to the equivalent widths can be of order of 30% for the weaker lines but certainly much lower for the stronger lines ( $\sim$ 5% for H $\alpha$ for instance). From Table 1 the mean radial velocities are: $v_{H{\sc I}}$ = 22 $\pm$ 14 kms^-1 (3 lines), $v_{He{\sc I}}$ = 67 $\pm$ 42 kms^-1 (3 lines), $v_{Fe{\sc I}}$ = 17 $\pm$ 7 kms^-1 (4 lines), $v_{Fe{\sc II}}$ = 13 $\pm$ 5 kms^-1 (5 lines) and $v_{O{\sc I}}$ = 7 $\pm$ 12 kms^-1 (2 lines). The errors reflect the root mean square of the averages. Several lines cannot be identified. Possible causes are line blending or existence of lines of high excitation ions. The radial velocities here reported were measured on the grism #7 spectrum, since this spectrum provided the best set of radial velocities, compared with the spectra taken with lower resolution grisms. For this reason we do not give velocities of lines with $\lambda$ > 6900 Å, since these measurements showed a great scatter. The above average of H I velocities does not consider the low velocity H $\delta$ and H $\epsilon$ lines, since these lines are likely contaminated with Fe I and Fe II and might suffer of wavelength calibration biases on the CCD blue edge. The velocities reported here are different from those given by Kilkenny (1997), viz. $v_{H{\sc I}} = - 6 \pm 10$ kms^-1 and $v_{Fe{\sc I}} = -28\pm 24$ kms^-1. A long-term study is necessary to confirm the radial velocity variability of this star.

3 Results