This work had as object the investigation and measurement of direct acoustic fields in shallow surface layers of the sea at short ranges (<1 mile in general) during summer when thermal conditions are complex and changeable; i.e., the correlation of experimental intensities with values computed from thermograms, after steps are taken to (1) minimize signal fluctuations due to transducer-rotations and swell, (2) ensure that direct field intensities only are accepted for analysis, (3) allow for thermal structure change in the relevant layers during recording, and (4) ensure for a few selected experiments that no change occurs in the vertical gradient along the transmission paths during recording. During six weeks in the 1954 summer, 18 experiments performed under calm conditions in depths of 30 to 200 fathoms and in areas about 20 miles from the New Zealand coast, yielded 30 000 values of acoustic pressure. In each experiment the field was sampled simultaneously at 6 depths from 40 to 165 ft, by a “string” of hydrophones, the sampling being repeated at short intervals from 100 up to 1700 yards. The frequency, repetition-rate, and pulse duration were respectively: 14.5 kc; 10 per minute; 1.3 msec. During an experiment the projector-depth was constant but values from 40 to 100 ft were used, depending on the temperature gradient encountered. This depth was chosen after inspection of the first of a number of thermograms taken per experiment, the criteria being (1) thermogram simplicity near the projector, (2) estimated relative absence of foci, (3) relative independence of the field upon swell, (4) estimated resolution of the direct and echo pulses at 1000 yards (or more), and (5) minimum downward refraction to ensure adequate signal-to-noise ratios at 1000 yards. These criteria conflict and demand compromise. From each relevant thermogram, intensity anomalies are calculated for points where analysis indicates successful resolution. It has been concluded that spatial thermogram constancy was best when both ships were allowed to drift with the currents. It has been found that (1) for a few optimum experiments (simple constant thermal structure and small swell) the agreement between the intensity and transmission anomalies was good and gave an attenuation at 14.5 kc of (2 to 4) db per kiloyard. (2) At small ranges (<250 yards) and for large differences in projector and hydrophone depths, large projector directivity correction factors (>3.0) give poor accuracy in the intensity and hence in the transmission anomaly. (3) For gradients producing marked downward refraction and for long ranges, both anomalies are large (10 to 20 db) and the measurement accuracy of the acoustic pressure is then poor. (Values <1.0 dynes per square centimeter.) (4) For the remainder of the 18 experiments, the thermogram-constancy was poor, the swell not negligible, and analysis showed the position of the projector in the gradient to have been not infrequently “unpromising.” In general, for such experiments wide variation and poor consistency occurred in the computed attenuation. The conclusion is reached that the discordant and variable values in attenuation obtained in past work at comparable frequencies have been largely the result of swell, of inconstant transducer geometry, of erratic changes in temperature gradient with time and space, and of incomplete resolution between direct and adjacent-path pulses.