The observations were made during seven consecutive nights from February,
28 to March, 06, 1989. We used a CCD camera attached to the Cassegrain
f/15 focus of the 1.0 m Jacobus Kapteyn Telescope, in the Roque de los
Muchachos Observatory, La Palma, Canary Islands. The detector used was
a GEC CCD 
pixels (1 pixel = 22 m)
and the pixel size is equivalent to 
on the sky. Typical readout
noise was 9 electons per pixel.
We performed narrow band imagery using interference filters centered on
the red shifted emission lines [OII]
, H
[OIII]
, H
, [SII]
and
[SIII]
, as well as on the continua at 3812 Å, 4556 Å,
5276 Å, 6269 Å and 9193 Å. In Table 1 (click here) we show the
main characteristics of the filters, the number of frames obtained in
each filter and the total integration time per filter. Throughout the
run the seeing ranged from 
to 
, reaching
exceptionally 
.
Table 1: Narrow band filters characteristics: Col. 1, ions and continua
identification; Cols. 2, 3 and 4 give respectively, the filter central
wavelength, the FWHM and the maximum transmission; Col. 5, rest
wavelength of the transmitted lines; Col. 6, filter transmission at
the wavelength of the line corrected by redshift and filter bandpass
thermal shift; Col. 7, number of co-added frames; Col. 8, total
integration times
Data reductions were carried out using a Silicon Graphics INDIGO
workstation, employing standard methods of the IRAF
reduction
package, removing bad pixels, cosmic
ray spots, subtracting bias, normalizing by flat field and correcting
atmospheric extinction. Four or more frames were co-added in each filter.
They were previously re-centered with a maximum error of 
pix and
convolved with a 2D-Gaussian function, in order to correct the seeing
differences.
The spectrophotometric standard HD84937 (Oke & Gunn 1983) was observed to calibrate the final images in absolute flux. This calibration and the subtraction of continuum frames from line plus continuum ones, to map the emission line distribution, were performed following the method outlined by Barth et al. (1994). The sky subtraction is automatically done in this procedure, provided that the sky is similar in both, the line plus continuum and the nearby continuum frames.
The sky background subtraction constitutes a difficult task to
study NGC4736 brightness distribution, since the CCD edges are
contaminated by light of the galaxy. In Table 2 (click here) we present
the sky brightness at each wavelength as measured in the galaxy and
standard star frames. In the galaxy frames we averaged four corner
areas of 
pixels and in the standard star frames about
half of the frame. Furthermore, in order to improve the S/N, the
background brightness from line plus continuum and continuum
frames for each line were also averaged. The behavior of the
background brightness as a function of the wavelength is the same in
both data sets, but it is systematically 40% more intense in the galaxy
frames than in the standard star ones. Then, to carry out the sky
subtraction in the surface photometry, we preferred the sky background
measured in the standard star frames.
We measured the emission lines absolute fluxes of the HII region of
NGC4736 by aperture photometry (see Sect. 3 (click here)). The galactic
extinction in the zone is E(B-V) = 0.0 (Burnstein & Heiles
1984). The HII regions internal reddening was obtained from the
decrement of the nebular H Balmer lines. Balmer emission lines intensity
was corrected of the underlying absorption line effect, produced mainly by
B, A, and F stars from the ionizing association. This correction has been
based on the average value of 1.9 Å of the H and H absorption
equivalent width, obtained by McCall et al. (1985) from a
sample of 99 HII regions in 20 spiral and irregular galaxies.
Furthermore, the H filter is transparent to the
[NII]
lines. This contamination has been estimated
with H /[NII]
=3 relative intensities quoted by
Burbidge & Burbidge (1962) and the filter transmission quoted
in Table 1 (click here). The correction amounts to 14% of the H\
intensity.
The sources of external errors of the observations were analyzed
following Barth et al. (1994). The Poissonian noise
for the integrated counts of the most frequent HII regions is
approximately 0.3% in H , 1.8% in H , 1% in [OIII] and [SII ],
and 3.4% in [OII] and [SIII].
For the atmospheric extinction we adopted the Roque de los Muchachos
Observatory's standard curve. The integration times by frame (e.g.
1800 s) imply a maximum variation in zenithal angle of 
during an exposure. Therefore, the use of a mean air mass do not
introduces large errors in the photometry.
The differences between the environmental temperature during the
observations and that of the laboratory at which the filter
transmission curve is measured (
C), cause a shift of 0.3
Å 
C-1 of the curve
maximum. This effect has been considered carefully in our reductions,
since the instrumental temperature has been registered during the
observations.
The statistical fluctuations of the counts in different combined frames
also introduce an error. For example, 18 independent observations in
H of the standard star give a mean of 22286 counts with a standard
deviation of 3.7%, after the correction for atmospheric extinction.
A reasonable estimate is that the errors introduced by the above
mentioned effects in the absolute flux of a typical HII region would
be about 6% in [OII], 5% in H and [OIII], 4% in H and 6%
in [SII] and [SIII].