A wideband interferometer is sensitive to the effects of longitudinal dispersion which affect the interfering light beams unequally. At shorter wavelengths the major effect of dispersion is from dry air itself, while at mid infrared wavelengths the effect of water vapor is dominant. MIDI, the future 10 micron instrument of the VLTI, will experience significant effects from the imbalances in the water vapor content affecting the paths of the two interfering beams. This imbalance will include terms due to the unbalanced air paths in the delay line, random atmospheric humidity fluctuations in the lines-of-sight to the star, and random humidity variations inside of the VLTI delay line tunnels.
Large amounts of dispersion, if not monitored, can reduce the accuracy of measured visibility amplitudes. Measurements of the visibility phase as a function of wavelength will be highly sensitive
to dispersion. This will then become a source of noise in results dependent on the phase of interference, such as imaging of non-symmetric objects, or detection of faint companions. In addition
to dispersion over the 7 - 14 micron region detected by MIDI, observations using phase tracking with detection in the near IR, could be catastrophically affected by differential phase delays between the 2 micron and 10 micron bands.
Dispersion measurements from VINCI observing in the K band, both due to dry air and to water vapor, are presented. Combining VINCI results with published data from millimeter wave measurements leads us to expect atmospheric differential water vapor fluctuations to exceed 1 mole/m2 rms over typical baselines. Specific effects from such a level of unmonitored dispersion variations are presented, which demand corrective action. Various solutions to monitor water vapor dispersion in realtime are