By automatically reassembling periodic samples of recorded speech, Fairbanks and associates have shown that comprehension is good for: (1) connected speech compressed 50% (0.5) in time, and (2) a single well‐learned PB list compressed up to 0.15. In 1919–1925 Fletcher and associates, using turntable speedup with its attendant frequency distortion, found that the intelligibility of syllables began to drop rapidly at speedups corresponding to time reductions of 0.8 (down about 25% at 0.7). In one case a complex technique using simple language showed only slight losses in comprehension for large time compressions. In the other case, a simple technique using difficult speech material showed marked deterioration. In our work on message (simple language) storage schemes, the desirability of obtaining time compressed speech with simple modifications of existing equipment has become evident. By merely speeding up the playback of a tape recorder, we have found that numbers and simple phrases can tolerate a time compression of 0.67 without serious losses in intelligibility, and PB words lose only 10%, in intelligibility with a time compression of 0.75.