Results

The neutral gas column density, compared with the B-band optical emission, is shown in Figure 1. There are several large gaps in the distribution, and a strong impression of multi-arm spiral structure. However these `arms' make no impression on the velocity field (Figure 2). Curiously, the sense of curvature in the HI is opposite to that of the stubby optical `arms' noted by e.g. Sandage & Bedke (1995). Despite the appearance of Figure 1, inspection of the channel maps and position-velocity cuts through the cube found that few of the holes in the disk could be clearly identified as supershells; this may be due to a combination of beam smearing and sensitivity limits.

Outside the central HI clumps, star formation is concentrated in the HI ridge running along the kinematic minor axis (PA) -- there is almost none associated with the other `arms'. Using the rotation curve derived below, I calculated that the critical density for gravitational instability was several times higher than the azimuthally averaged HI column density, at all radii. However in the HI ridge, the surface density is within a factor of 2 of the critical value. Considering the uncertainties and lack of data on the distribution of the molecular gas component, it seems possible the outer star formation regions NGC 4214 can be explained by gravitational instability.

The striking features of the velocity field (Figure 2) are the strong decline in the projected velocity at large radii, and the twist of the isovelocity contours in the central regions. As discussed by Allsopp (1979), there are a number of ways in which such a velocity field could be produced. A warp of the disk to lower inclination at large galactocentric radius seems quite likely, given that deep CCD images (McIntyre 1996, in prep) show increasingly circular isophotes in the region where the rotation curve is falling. How much of a warp is required depends on the inclination of the galaxy as a whole, which is uncertain. For an inclination of 30, the inclination of the outer disk must decrease by 15 over the outer 200.

However a warped-disk model cannot explain the strong central isovelocity twist. I have attempted to model the velocity field as a disk with oval orbits in the central regions, that smoothly become more circular with increasing galactocentric radius. The orbit shapes are parameterised in a way similar to that of Staveley-Smith et al. (1990). I took the inclination to be constant at all radii and fitted the shape of the rotation curve with a Brandt (1960) parameterisation. This assumption doesn't significantly affect the fit to the central regions, as the region of interest is well within the turnover radius. Since it is rather low, the inclination was not fitted automatically; several fixed values were tested. All the free parameters were fitted together, using a simulated annealing algorithm adapted from Press et al. (1992) to minimise the rms difference between the computed and observed velocity field. The best fits were for orbits with rather low ellipticity () with major axes along PA, roughly parallel to the long axis of the bright optical bar. The orbits were significantly oval only for . The fitted rotation curve has a Brandt n parameter of , and peaks at  km , 160'' from the rotation centre.

 
Figure: Total HI column density maps of NGC 4214, at 30'' (left) and 8'' (right) resolution. To show maximum detail, different greyscales have been applied to each image. The contours show the outline of the galaxy in the Johnson B band; field stars are marked with small crosses, the rotation centre with a large cross.

 
Figure 2: The velocity field of NGC 4214, at 30'' resolution. The cross marks the rotation centre obtained from the model fit described in the text.

© Copyright Astronomical Society of Australia 1997