We have decomposed *J*- and *K*-band images of 14 early-type
spirals into bulge and disk components.
2D non-parametric solutions and results from fitting a parametric
model of generalized exponential 1/*n* bulges and simple exponential disks
are compared, and general characteristics of early-type spiral bulges
and disks are examined.
We find that:

- 1.
- Even using objective and refined techniques, the decomposition in structural components is far from being a robust and unique process. For the parametric methods, significantly different decompositions are obtained for different bulge distribution laws. Non-parametric techniques, on the other hand, appear to be affected by the choice of the ellipticities of the components which are difficult to evaluate objectively. Of the two components, the bulge is the most subject to errors, since the inner part is masked by seeing, and the outer regions are buried beneath the disk.
- 2.
- Bulge structural parameters are strongly influenced by the
form of the function used to derive them.
The same bulge, when fitted with small
*n*, appears to be ``denser'' (brighter ), more compact (smaller ), and less luminous than when fitted with large*n*. The dispersion of the fitted parameters also increases with*n*. - 3.
- The median early-type bulge has a shape index
*n*between 2 and 3, =16.8mag arcsec^{-2}, and = 1.6 kpc. It is also red, with = 1.06, and redder bulges tend to be ``denser'', that is with brighter . - 4.
- As noted by Kent (1988), bulges are rarely spherical. The median intrinsic ellipticity is 0.34, equivalent to a disk with 50 declination. This restricts the applicability of non-parametric techniques to rather inclined systems and suggests treating with caution the studies which assume spherical bulges.
- 5.
- The median early-type disk with = 17.1
is more than 1 mag arcsec
^{-2}brighter than later-type disks, and bluer than the bulge in (*J*-*K*) by more than 0.1 mag. Disk scale lengths agree fairly well with those found by other at different wavelengths, and we confirm a tendency for NIR disk scale lengths to be smaller than those at optical wavelengths (e.g., Peletier et al. 1994). is approximately constant, 0.24, similar to the value of 0.25 found for late-type spiral disks (Giovanardi & Hunt 1988; Giovanelli et al. 1995). - 6.
- Both bulge and disk surface brightnesses correlate with respective
scale lengths, consistently with the projection of the
fundamental plane for ellipticals and spiral bulges
(e.g., Andredakis et al. 1995).
We note that uncertainties in the decomposition, especially
in the shape index
*n*, strongly influence the position of a bulge within the FP. Disks appear to reside in a region of this FP projection which is roughly contiguous to that of bulges, extending the correlation to larger radii and fainter surface brightnesses. - 7.
- We confirm the tendency
for the ratio of bulge and disk scale lengths to be constant,
noted by de Jong (1996a) and
Courteau et al. (1996).
However, we find a mean (best-
*n*) value = 0.3, significantly larger than the value found by de Jong and Courteau et al.; our*n*= 1 value of of 0.2 agrees roughly with their value of 0.13-0.14, while our*n*= 4 value is 0.7, more than a factor of 3 larger. We attribute such differences to different bulge parameterizations and caution that if best*n*varies with morphological type, as suggested by Andredakis et al. (1995), may not be constant with morphological type, and thus the Hubble sequence may not be scale free as proposed by Courteau et al. (1996).

We would like to thank the referee, F. Simien, for a thorough reading of the manuscript and useful suggestions. This research was partially funded by ASI Grant 95-RS-120.

Copyright The European Southern Observatory (ESO)