V. I. Tatarski, Wave Propagation in a Turbulent Medium (McGraw-Hill Book Company, Inc., New York, 1961), Chap. 13.

As a measure of aperture averaging Tatarski computes a quantity G, defined by his Eq. (13.28), with results presented in his Fig. 36. The equation defining G can be cast in the form [equation] where *D* is the aperture diameter, σ^{2} is the variance of the logintensity, and *b*_{A}(*D**x*) is what we have called the normalized log-amplitude covariance function [which we have denoted by *C*_{l}(*Dx*)/*C*_{l}(0)]. (The factor 1/σ^{2} which Tatarski has in front of the logarithm in Eq. (13.28) is absorbed into the integrand in our expression by the minus one in the exponent. The minus one is absent in Tatarski's expression.) We need only note that σ is non-negative and that [*b*_{A} (*Dx*) - 1] is non-positive for all values of D and x. The exact physical significance of *b*_{A} and σ need not concern us. The derivative of G with respect to σ is [equation] The integrand in the numerator is everywhere negative, while the integrand in the denominator is everywhere non-negative. Obviously, ∂*G*/∂σ is negative. For any particular value of D, the value of G should be decreased when σ is increased. This condition is clearly contradicted by Tatarski's Fig. 36.

The shot noise σ_{N} is intended to refer to that (quantum) noise which would be present if there were no atmospherically induced irradiance fluctuations. Even a modestly sophisticated system can avoid confusing the atmospherically induced irradiance fluctuations with the signal variations which are meant to carry information. Simple voice-controlled intensity modulation at the transmitter, of course, results in the atmospherically induced signal variations appearing as noise, as has been considered to be the case by J. I. Davis, Appl. Opt. 5, 139 (1966), but binary coding, or AM—FM modulation permits voice information to be transmitted without the atmospheric effects coming through as a spurious signal, i.e., as noise. Only the quantum fluctuations cannot be eliminated. σ_{N} refers only to this noise.

D. L. Fried and J. D. Cloud, J. Opt. Soc. Am. 56, 1667 (1966).

The log-amplitude *l*(*x*), normally computed, equals the logarithm of the ratio of the amplitude measured to the amplitude which would be measured if there were no refractive-index variation in the propagation path. Because energy must be conserved, and refractive-index variations do not dissipate optical energy, in the propagation of an infinite plane wave or a spherical wave, the measured average irradiance, *I*_{0}, corresponds to the irradiance which would be observed in the absence of atmospheric turbulence. For infinite-plane-wave propagation, the amplitude computed from the square root of *I*_{0} can be related to the amplitude which would exist in the absence of turbulence. For propagation of a collimated beam, it cannot. Thus, if we wished to apply) Eq. (1.1) to propagation of a laser beam, I_{0} could not be interpreted as the measured average irradiance, but rather as the irradiance that would be measured if there were no turbulence.

Ref. 1, p. 209 and p. 210.

G. E. Mevers, *el al.*, J. Opt. Soc. Am. 55, 1575A (1965).

The quantity (4*z*/*k*)½ is a unit of length which arises in the analysis of the log-amplitude statistics for horizontal propagation of an infinite plane wave. It is useful in normalizing correlation distance so that results can be expressed in nondimensional form, as is developed in Ref. 4. Tatarski uses (λ*z*)½ for this same purpose. For propagation from space to the ground, the normalization distance is (4*h*_{0}secθ/*k*)½ where θ is the zenith angle of the apparent source, *k* is the wave number, and *h*_{0}= 3200 m is a quantity which characterizes the vertical distribution of atmospheric turbulence.

D. L. Fried, J. Opt. Soc. Am. 55, 1427 (1965).