On July 20, 1962 the National Bureau of Standards issued Experimental Transition Probabilities for Spectral Lines of Seventy Elements by Charles H. Corliss and William R. Bozman.
Section 9 of the Introduction, "Computation and Preparation of the Tables" (p. XVI) reads as follows:
"At the beginning of the preparation of this table it was realized that equipment was available which would permit essentially automatic preparation of the finished book. It was therefore decided to attempt to produce this publication by completely automatic methods. An electronic computer could be used for the computation, then the magnetic tape output from the computer could be used to operate an automatic phototypesetting machine which would produce film ready for making the printing plates.
"The basic input data for the computation were already available on a magnetic tape, having been prepared previously for the NBS Tables of Spectral-Line Intensities. The wavelength, element name, spectrum assignment, and the two energy levels for each classified line were selected from this tape. The other parameters in eq (9) and (10) for each element and spectrum were read into the computer from punched cards.
"The computation could have been done in one step, with the output tape written in the form for use by the phototypesetting machine; however, for purposes of checking, it was decided first to write the output data for printing on the off-line computer, then as a separate operation to convert this Binary-Coded-Decimal tape to a binary tape in the proper format for the phototypesetting machine. The computer instruction coding for the first step was written entirely in the FORTRAN (FORmula TRANslating) automatic coding system. The coding for the second step was written mostly in FORTRAN, but the actual convert portion was written as a SUBROUTINE sub-program in FAP machine language. The first step required about 20 min running time, and the second step about 13 min on a 7090 computer. The resulting binary tape, written in 256- character records, had to be rewritten onto 30 tapes of two-character records in order to match the input buffer. The pages were photographically typeset with a Mergenthaler Linofilm machine onto 8-in. wide film in 50-ft rolls. About 8 min were required to set each page. Printing plates were made from the film positives which are actual page size."
This book may also be unique, or at least extremely unusual for a Government Printing Office publication, in having a colophon. It reads:
"This is the first production-line book in which the tables were composed by a photocomposition machine controlled by the output of a digital computer. Printing instructions, column headings, and the decimal tabular material were programmed for the computer. The output magnetic tape then became the input to the photocomposition machine which produced auto-positive film. These, in turn, were used to produce direct offset printing plates."