Data from the Palomar Testbed Interferometer , with a baseline length of 110 m and an observing wavelength of 2.2 \mu { m } , were used to derive information on atmospheric turbulence on 64 nights in 1999 .
The measured two-aperture variance coherence times at 2.2 \mu { m } ranged from 25 msec to 415 msec ( the lower value was set by instrumental limitations—the interferometer could not operate when the coherence time was lower than this ) .
On all nights , the spectrum of the short time scale ( < 600 msec ) delay fluctuations had a shallower spectrum than the theoretical Kolmogorov value of 5/3 .
On most nights , the mean value of the power law slope was between 1.40 and 1.50 .
Such a sub-Kolmogorov slope will result in the seeing improving as the \approx 0.4 power of wavelength , rather than the slower 0.2 power predicted by Kolmogorov theory .

On four nights , the combination of delay and angle tracking measurements allowed a derivation of the ( multiple ) wind velocities of the turbulent layers , for a frozen-flow model .
The derived wind velocities were all \leq 4 { m s ^ { -1 } } , except for a small 10 { m s ^ { -1 } } component on one night .

The combination of measured coherence time , turbulence spectral slope , and wind velocity for the turbulent layer ( s ) allowed a robust solution for the outer scale size ( beyond which the fluctuations do not increase ) .
On the four nights with angle tracking data , the outer scale varied from 6 to 54 m , with most values in the 10–25 m range .
Such small outer scale values cause some components of visibility and astrometric errors to average down rapidly .