Astron. Astrophys. 360, 85-91 (2000)

2. The revision of literature data

2.1. Description

The data from the Fal89, Pal90 and Gal94 surveys were obtained in machine readable form from files kindly provided by the authors. Those by Gal94 may also be obtained from CDS, Strasbourg. The data files were entered into ad hoc MIDAS procedures, allowing the following operations:

1. check the run of colour against and eventually introduce small "corrections" to the adopted sky-background levels, in order to obtain a nice linear plot (if feasible!). This exercize is conducted simultaneously upon the 3 passbands (and colours) available in each dataset, to minimize the introduced corrections.

2. calculate synthetic images and calibrate these against the same sources as our own, primarily Poulain's UBVRI photometry (Poulain 1988; Poulain & Nieto 1994), in order to place all data in an homogeneous system, with UBV in Johnson's and RI in Cousins's systems.

Remark: Measurements from BM87 for 7 Virgo objects were incorporated in the present discussion. Gradients in V-I, or R-I, were recovered from the published graphs. The V-I colours at were found from a correlation between this quantity and the mean colours within given by Poulain in his surveys: these two colours are equal within errors.

2.2. Short discussion

2.2.1. Changes in background values and colour gradients

As the observations in Fal89 and Pal90 were performed largely with a small telescope and a relatively large field, their sky-background estimates were generally good enough to require little ad hoc "corrections". Thus we treated 29 files from Fal89: the changes in background were less than 1% in 3/4 of the cases and reached 2% in only two cases. For Pal90, the corresponding figures for 51 files treated are nearly the same. The data in Gal94 were obtained with cameras having low fields of view, and, notwithstanding the care of the authors in recovering good background values, we were led to more frequent and somewhat larger changes. We treated 88 files: the "corrections" were less than 1% in 70% of cases but were larger than 2% in 8 cases.

These so called "corrections" to the sky-background values adopted by the observers may appear arbitrary, since there is no obvious physical reason for the colours in E galaxies to remain linear in up to the largest r values. They might instead tend towards definite limits at large r, when stellar populations reach some average, and dust is nearly gone. We assume that the present observations do not reach this asymptotic region. In Fig. 1 we compare the correlation diagrams of the B-V and B-I gradients obtained from the Gal94 data before and after the revision of the background values. The expected correlation is clearly improved by our "corrections", which is an a posteriori justification for our procedure. On the other hand, the correlation between the B-R and U-R gradients is not good, as already noted in Pal90, and is barely improved by our revision of background values.

Fig. 1. The correlations between the gradients in B-I and B-V from Gal94 data before (upper graph) and after (lower graph) the described revision. Our "corrections" clearly improves the correlation.

The new derived gradients may differ from the original ones for other reasons than changes to the background values, because we sometimes modified also the inner radius limit in the gradient calculations. First, when the object was known to contain an important inner dust pattern the heavily affected region was left out of the fit: this applies to NGC1052, 2768, 2974, 3665, 4278, 4374. Second we consider that Gal94 were not stringent enough in applying inner cutoffs to the colour data before the least square fit. Judging from the appearance of the colour graphs we often used a more restricted fitting range.

The newly derived gradients do not differ much from the original. A little statistic of the New-Old values give the following results:

. Fal89 and Pal90 data together:
N=22 mean
N=21 mean

. Gal94 data:
N=25 mean=-.002
N=21 mean=.005 .
The objects with large dust patterns are rejected from this last comparison.

2.2.2. New calibrations

The possible improvement in calibrations arise mainly from the availability of the above quoted Poulain's aperture photometry. A comparison of New-Old colours, taken of course at the same radii and before the usual corrections for K-effect and galactic extinction, gives the following results:

. Fal89: B-R, 9 cases, mean -0.295, a large mean difference due to the use of Johnson's R in Fal89.
U-B, 8 cases, mean -0.021,

. Pal90: B-R, 13 cases, mean -0.053,
U-B, 13 cases, mean -0.025,

. Gal94: B-V, 19 cases, mean -0.015,
V-I, 16 cases, mean 0.040,
again indicating that the I photometry of Gal94 contains a number of poor calibrations, leading to several unrealistic V-I (see also Michard 1999).

While the present new colour calibration is more homogeneous it does not necessarily leads to smaller random errors than the original ones. It should be noted that the data files used for the Fal89 and Pal90 surveys are truncated in B and U to some inner radius of several arcsec, corresponding to the limit where the colours were thought to be reliable by the authors. To build synthetic images for the new calibrations, this inner range had to be extrapolated, a procedure likely to introduce errors in the results. We shall come back below to this question.

© European Southern Observatory (ESO) 2000

Online publication: July 27, 2000
helpdesk.link@springer.de