3 Description of the spectra

Figure 1: The continuum normalized FeII 5169 Å line before the light maximum. Three absorption features shifted by -1100, -1700 and -2200 km s-1 are marked

The first two spectra were obtained just before the light maximum, which occured on May 5.0 UT (JD 2451669.5, see later). The bluer one was centered at 5184 Å and showed the strong Fe II 5169 Å emission line with P-Cyg profile (Fig. 1). The radial velocities of absorption features are -1100, -1700 and -2200 km s$^{\rm -1}$. (This wavelength range was observed again on two other nights, but the acquired spectra had too poor S/N ratios for further conclusions.) The simultaneously observed H$\alpha$line is nearly Gaussian with slight redward asymmetry (top curve in Fig 6). Note the presence of the diffuse interstellar band (DIB) at 6613 Å. The later evolution of the H$\alpha$ line will be described separately, thus we bring forward discussion of other spectra.

Figure 2: The intermediate resolution spectrum of CI Aql in the blue region

One intermediate resolution spectrum was taken on May 16.3 UT centered at 4200 Å ($\Delta t$ = +11 d). Prominent Balmer emission lines with strong P-Cyg profiles dominate the spectrum, while Fe II 4176 and 4233 Å lines are also present (Fig. 2). An H$\alpha$ observation on the same night reveals the similarity of all observed hydrogen lines: double-structured absorption suggesting two expanding shells with velocities of -1800 and -2400 km s$^{\rm -1}$ (Fig. 3).

Figure 3: Medium resolution hydrogen profiles. Two well-defined absorptions are present at -1800 and -2400 km s-1. The broad windgs extend to $\pm$3000-4000 km s-1. The continuum normalized spectra were shifted by 0.5 for clarity

Figure 4: The He I 5876 and Na I D blend. Note the presence of a strong interstellar component implying fairly high reddening

The next run was on May 25 ($\Delta t$ = +21 d), when the He I 5876 Å/Na I D blend was observed. The most important detection in this spectrum is of the strong interstellar component (see the deep and sharp features at the rest wavelengths of sodium doublet in Fig. 4) and the much weaker DIB at 5849 Å. These lines can be used as reddening indicators as will be discussed in the next section.

Figure 5: A series of low-resolution objective prism spectra for CI Aql. A vertically shift of 2 was applied to avoid overplotting

The first low-resolution objective prism spectrum taken a day later is shown in Fig. 5. We could identify the following lines/blends: hydrogen Balmer series from H$\alpha$ to H$\gamma$, N III 4640/He II 4686, N III 5001/He I 5016, N II 5679, He I 5876/Na I D, He I 7075, He I 7281 and He II 8237 Å. Further objective prism spectra taken in the next four days did not show significant variations, except some changes in the He II lines, while the N III/He II blend became stronger than H$\beta$. All of these features are typical for a "He/N'' nova during the permitted phase (see Fig. 2 in Williams 1992). Additionally, we took a 3-hours long time-series objective prism observation on May 30, but did not find significant changes.

Finally, the largest number of spectra addressed the evolution of the H$\alpha$ line. Being the strongest emission line, it could be observed even in the fainter state in late June. Figure 6 summarizes the line profile variations. The early quasi-Gaussian profile changed to a P-Cyg profile implying expansion velocities above 2000 km s^-1. The broad wings extend to $\pm$3000-4000 km s^-1 suggesting an early FWZI of $\sim$8000 km s^-1. The detected 200 Å wide region prevented obtaining a proper FWZI-curve as has been presented for U Sco by Munari et al. (1999) and Anupama & Dewangan (2000), but parallel to the shape variations, the H$\alpha$ line became slightly narrower. The strongest line profile change occured between day +21 and +33, when the saddle-shaped profile formed with two maxima at $\pm$1100 km s^-1.

We have also tried to find short-term variations in the line profile by taking 8 spectra in a 3-hours long period on June 6, but the attempt has failed.

The presented spectra will be compared with other RNe in Sect. 6, here we only mention that the observed behavior is very similar to other U Sco-type RNe (U Sco, V394 CrA), only the time scales are different.

Figure 6: The evolution of the H$\alpha$ profile during the first two months after the maximum. The continuum spectra were re-normalized at 6586 Å (V=1100 km s-1) in order to enable an easy comparison of the line shapes (the continuum normalized flux maxima ranged between 6 and 150). Each subsequent spectrum is shifted upwards by 0.5