For this analysis we need data to provide us with information on the stellar mass distribution and on the gas kinematics of a sample of galaxies. To map the stellar surface mass density it is most desirable to take near infrared (NIR) images of the galaxies, because dust extinction and population effects are minimized (e.g. Rix, Rieke, 1993). During two observing runs in May '99 and March '00 we obtained photometric data for ~20 closeby NGC galaxies. We used the Omega Prime camera at the Calar Alto 3.5m telescope with the K-band filter (K'). It provides us with a field of view of 6.76' x 6.76'. Figure 1 shows the K-band image of the Messier galaxy M99.
The kinematic data was obtained with the TWIN, a longslit spectrograph for the 3.5m telescope. To reach a reasonable coverage of the galaxies' velocity field, we needed to take 8 slit positions across the entire disk of the galaxies (also displayed in Figure 1). So far we were able to collect complete sets of longslit spectra for only 4 galaxies, mostly due to only moderate weather conditions during the spectroscopy runs.

As a pilot project, we analyzed the data of NGC 4254 (M99). Assuming a constant stellar mass to light ratio, the gravitational potential due to the stellar mass fraction was calculated by direct integration over the whole mass distribution taken from the NIR-image. The mass to light ratio for the maximum disk contribution was scaled by the measured rotation curve. For the dark matter contribution we assumed an isothermal halo with a core. To combine the two components we chose a stellar mass fraction and added the halo with the variable parameters adjusted to give a best fit to the rotation curve.
We used this potential as an input for the hydrodynamical gas simulations. Figure 2 presents the results for the resulting gas surface density, as it settles in the potenital. The morphology of the gas distribution is very sensitive to the velocity, with which the spiral pattern of the galaxy rotates (pattern speed). In figure 2 we printed the result of the simulation with the spiral structure matching best to the K-band image morphology. We find quite good agreement.
In figure 3 a comparison of the modelled and the measured rotation curves are presented for one particular position angle. The four panels show the rotation curves for different disk mass fractions, ranging from almost none to maximum disk case.

Although there is quite some scatter in the observed data, we find that the velocity jumps, which are apparent in the simulations for the 85% case are too large to be in agreement with the measurements. On the other hand the simulated velocity wiggles in the 20% case seem too small to match the observations. The inner part of the simulated rotation curve (< 0.5') is dominated by the dynamics of the small bar, which is present at the center of the galaxy. Its pattern speed might be different from the one of the spirals and thus relate to a mismatch in the inner part of the rotation curve.
We conclude that an axisymmetric dark halo is needed to explain the kinematics of the stellar disk. The influence of the stellar disk is submaximal in that respect that we don't find strong enough velocity wiggles in the observed kinematics as it would be expected if the stellar disk was the major gravitating source inside the inner few disk scale lenghts.
How this conclusion might apply to other spiral galaxies will be the upcoming issue of this project. We plan to extend our analysis at first on the 3 other galaxies where we have already now complete data sets. Finally we intend to draw our final conclusions on a basis of a sample consisting of 8 - 10 members. This should be sufficient to determine reliable results about the luminous and dark mass distributions in spiral galaxies.