2 Observations and data reduction

All observations were made with the ESO coudé auxiliary telescope (CAT) and the coudé echelle spectrograph (CES) during the period November 18-24, 1992. We used a $1024\times1024$ pixel CCD at a resolving power of 50000 (spectral resolution of $\approx$0.12 Å). The integration time of the spectra was between 1200 and 3600s, which allowed for an average signal-to-noise ratio of 70:1 for UX For and 150:1 for AG Dor. Each spectrum covers $\approx$50 Å from $\lambda$6526 Å to $\lambda$6576 Å. A typical night of observations includes two biases, 10 flat-field images, 5-7 thorium-argon lamp integrations at the beginning and at the end of each night, and 1-3 integrations of a rapidly rotating B-star (for telluric line correction).

All data were reduced in the same standard fashion using the Image Reduction and Analysis Facility (IRAF) and included bias subtraction, flat fielding, optimal aperture extraction, and telluric line division. The flat-field exposures were averaged together and used to remove the pixel-to-pixel variation of the detector sensitivity. Th-Ar comparison spectra and spectra of a bright radial-velocity standard star were obtained at least once during each night to ensure an accurate wavelength calibration.

A total of 37 spectra of UX For (see Table 1) and 38 spectra of AG Dor (Table 2) were obtained. Additionally, a total of 14 bright-spectrum comparison stars were observed throughout the run. We chose stars with well established spectral classifications (if possible MK standards) and with spectral lines not significantly broadened by rotation.

H$\alpha$ spectra are usually contaminated with many small absorption lines due to water vapor in the terrestrial atmosphere. Spectra of rapidly rotating hot stars show mostly a flat continuum near H$\alpha$ and are ideally suited to reconstruct the nightly telluric spectrum. We chose the rapidly rotating B star HR674 (spectral type: B8V-IV, $v\sin i\approx 250$ kms^-1). Several nightly spectra were combined to increase the signal-to-noise ratio and eliminate cosmic-ray hits. High-frequency noise was eliminated by setting all intensity values of the normalized B-star spectrum above 0.985 to unity. The remaining spectra, weighted by airmass, were then used for dividing the object spectra.

 

 

Table 1: Log of the observations of UX Fornacis and radial velocities

UT Date	HJD	phase	$v_{\rm r}^{\rm P}$	$v_{\rm r}^{\rm S}$	$\Delta t$
(1992)	(24489+)		[km s-1]	[km s-1]	[s]
Nov 19	45.5423	0.3221	-20.3	89.6	3600
19	45.7263	0.5148	-76.9	151.9	1800
19	45.7492	0.5387	-73.9	147.5	1800
19	45.7694	0.5599	-70.1	142.8	1200
19	45.8388	0.6326	-45.7	104.5	1200
19	45.8532	0.6477	-40.3	99.4	1050
20	46.5061	0.3315	-29.2	85.7	1200
20	46.5249	0.3512	-39.7	95.3	1200
20	46.5401	0.3671	-44.8	109.1	1200
20	46.5547	0.3824	-51.3	114.2	1200
20	46.5700	0.3984	-58.1	123.8	1200
20	46.5846	0.4137	-62.3	133.7	1200
20	46.5998	0.4296	-67.9	140.8	1200
20	46.6151	0.4456	-70.9	143.9	1200
20	46.6582	0.4908	-77.0	151.7	1200
20	46.7269	0.5627	-72.7	140.9	1200
20	46.7422	0.5787	-66.1	140.0	1200
20	46.7575	0.5948	-61.7	134.9	1200
20	46.7721	0.6101	-56.7	127.3	1200
20	46.7873	0.6260	-51.0	106.1	1200
20	46.8026	0.6420	-44.1	98.2	1200
20	46.8172	0.6573	-33.8	94.2	1200
20	46.8325	0.6733	-27.4	88.9	1200
20	46.8477	0.6892	-18.8	82.9	1200
21	47.5368	0.4109	-60.4	129.0	1200
21	47.5519	0.4268	-65.8	135.0	1200
21	47.8242	0.7119	-$\ \ \$	-$\ \ \$	1080
21	47.8377	0.7261	-$\ \ \$	-$\ \ \$	1140
21	47.8493	0.7382	-$\ \ \$	-$\ \ \$	720
22	48.5365	0.4580	-71.4	141.6	1200
22	48.5525	0.4747	-71.6	143.8	1200
23	49.7678	0.7475	-$\ \ \$	-$\ \ \$	1500
23	49.7862	0.7668	-$\ \ \$	-$\ \ \$	1500
23	49.8042	0.7857	-$\ \ \$	-$\ \ \$	1320
24	50.7710	0.7982	-$\ \ \$	-$\ \ \$	1200
24	50.7873	0.8153	64.2	-45.1	1500
24	50.8061	0.8350	73.4	-44.0	1500

 

 

Table 2: Observing log of AG Doradus and radial velocities

UT Date	HJD	phase	$v_{\rm r}^{\rm P}$	$v_{\rm r}^{\rm S,H_{\alpha} res}$	$\Delta t$
(1992)	(24489+)		[km s-1]	[km s-1]	[s]
Nov 19	45.5867	0.0354	124.5	-26.9	3600
19	45.7944	0.1165	111.8	-4.9	1800
19	45.8173	0.1254	110.1	-2.4	1800
20	46.6783	0.4615	14.9	171.6	1800
20	46.6998	0.4699	14.1	171.7	1800
21	47.5712	0.8100	86.0	37.1	1800
21	47.5933	0.8186	88.9	20.3	1800
21	47.6413	0.8374	95.1	23.4	1800
21	47.6633	0.8459	98.3	22.5	1800
21	47.7161	0.8665	104.6	15.3	1800
21	47.7381	0.8751	107.1	11.3	1800
21	47.7818	0.8922	111.5	5.6	1800
21	47.8037	0.9007	113.8	3.3	1800
22	48.5740	0.2014	87.7	20.1	1800
22	48.5955	0.2098	84.8	20.4	1800
22	48.6337	0.2247	79.9	26.0	1800
22	48.6552	0.2331	77.3	37.2	1800
22	48.6774	0.2418	74.1	48.1	1800
22	48.6990	0.2502	71.2	(7.4)	1800
22	48.7337	0.2637	66.6	55.2	1800
22	48.7559	0.2724	63.6	63.9	1800
22	48.7788	0.2813	60.4	(34.5)	1800
22	48.8274	0.3003	54.1	(44.7)	1800
22	48.8497	0.3090	51.5	(31.8)	1800
23	49.5635	0.5876	20.4	163.1	1800
23	49.5857	0.5963	21.6	159.3	1800
23	49.6278	0.6127	24.8	149.7	2400
23	49.6567	0.6240	26.5	148.2	2400
23	49.6961	0.6394	31.3	139.6	2100
23	49.7216	0.6493	33.6	134.3	2100
23	49.7456	0.6587	36.1	134.3	1800
23	49.8257	0.6900	45.9	129.8	2100
23	49.8512	0.6999	48.9	109.3	2100
24	50.5627	0.9776	124.7	-20.7	3600
24	50.6190	0.9996	125.5	-24.2	3600
24	50.6655	0.0178	125.5	-22.7	3600
24	50.7415	0.0474	123.9	-21.4	3600
24	50.8384	0.0852	118.7	-14.2	3480

Figure 1: A typical spectrum of AG Dor (thick line). The thin line shows the K0V standard star HR857. Note the blue-shifted H$\alpha$-emission line from the secondary component and the filling of the primary-star line due to chromospheric emission. Also shown is the spectral line used for Doppler imaging the primary component at 6546.24 Å along with four blends