A history of the Cross Section Evaluation Working Group is available here.
The following is taken from the ENDF/B-VII format manual, Appendix I "Historic Perspective", By Norman Holden.
The process of digesting experimental data, combining it with the predictions of nuclear model calculations and attempting to extract the true value of a cross section is referred to as an evaluation.
Historically, individual laboratories around the world had prepared evaluated nuclear data, e.g., neutron-cross sections, for their use in nuclear reactor calculations. These laboratories stressed their own needs for materials, cross section types and energy ranges depending upon their specific applications. Each of these laboratories developed their own methods for the storage and retrieval of these data.
In addition, it was noted that some neutron transport programs had built-in neutron cross-section libraries that could not be modified. As a result, reactor designers could not use new cross-section data, which in some cases had been available for more than half a decade.
There were fairly detailed nuclear data libraries available by 1963, the United Kingdom Nuclear Data Library (UKNDL) from Ken Parker at the Atomic Weapons Research Establishment, Atomic Energy Authority in Aldermaston, UK, the fast reactor data library from Joe Schmidt at the Institute for Neutron Physics and Reactor Technology, Nuclear Research Center, Karlsruhe, Germany, the NDA library from Herb Goldstein at Nuclear Development Associates, in New York, and the Evaluated Nuclear Data Library (ENDL) from Bob Howerton at the Lawrence Radiation Laboratory in Livermore, California to name just a few.
At the 1961 Vienna Conference on the Physics of Fast and Intermediate Reactors, Ken Parker indicated some of the requirements for the neutron cross section libraries. They had to specify reaction processes available or else a zero value cross section would automat- ically be assumed. There had to be a simple presentation of the data on punched cards, which would be easy to revise. However, the data could not be revised frequently or the reactor designers would be unable to perform comparative calculations, as they made design revisions. There was a need to cross check the data for errors and the best data should provide reasonable answers for simple systems, such as bare reactor cores.
At the 1964 Geneva Conference on Peaceful Uses of Atomic Energy, John Story from the Atomic Energy Establishment, Winfrith, UK, defined a data file as a complete set of evaluated cross section data for a single material and a data library as data files for a number of materials.
The various data libraries that were available often gave different answers, when the libraries were used to calculate the same reactor configuration. However, dissimilarities in the internal formats of the various libraries made it difficult to understand why these differences occurred.
There was a need for a common file between these existing systems, which would allow for an inter-comparison of these libraries. The stimulus for action came from a discussion among Henry Honeck of Brookhaven National Laboratory, Al Henry of Westinghouse and George Joanou of General Atomics at the Colony Restaurant in Washington, D.C. The Reactor Mathematics and Computation (RMC) Division of the American Nuclear Society (ANS) was requested to sponsor two meetings to discuss this common link, as a result of the above discussion. Honeck as chairman of the Division's sub-committee on Evaluated Nuclear Data Files held some meetings. A group of eighteen representatives from fifteen US laboratories met in New York City on July 19, 1963 to review cross section libraries and discuss means for interchanging these libraries. A sub-committee was appointed to meet in Hanford on September 18-20, 1963 to examine library formats in more detail.
The conclusions of these discussions were that there was a need for a standard format for evaluated nuclear data and the format should be as flexible as possible so that existing libraries could be translated into the standard format and that future needs could be easily incorporated into the file. This standard format would serve as a link between a data library and the processing codes. It was also suggested that a center should be established and charged with the development and the maintenance of the Evaluated Nuclear Data File (ENDF) and with the collection and distribution of data.
A preliminary report of the detailed formats for ENDF was sent for review and comment. Twenty-two people attended a final meeting at Brookhaven on May 4-5, 1964 to discuss changes and settle on a final version. The description of this system (which was labeled Version A and referred to as ENDF/A) was documented in the report BNL-8381. The ENDF/A file originally contained an updated version of the UKNDL library as well as evaluated data from a number of different laboratories.
The reactor designer wants evaluated data for all neutron-induced reactions covering the full range of incident neutron energies for each material in a reactor. However, evaluators usually supply "bits and pieces", which are put together to form a fully evaluated set for a given material. ENDF/A provides a storage system for these "bits and pieces" or partial evaluations. In addition to the need to allow all nuclear data evaluations to be placed on a common basis, there was also a need for an evaluated nuclear data file to be used for reactor design calculations. The description of this system (which was labeled Version B and referred to as ENDF/B) was documented in the report BNL-50066.
Where the format of ENDF/A was highly flexible in order to accept data in almost any arrangement or representation, the format of ENDF/B had to be simple to facilitate the writing of processing programs to use the data. This new ENDF/B library format would be mathematically rigorous, with specific interpolation schemes between tabulated points, so that cross section integrals, products and ratios would yield well-defined and repeatable results. There would be codes developed for plotting, integration and other processing of cross sections that would be written in FORTRAN for computer interchangeability and distributed to assist others who wanted to use ENDF data.
A material was defined as either an isotope or a collection of isotopes with a material number designated by the symbol MAT. The data for a material is divided into files with the file number designated by the symbol MF. A file is subdivided into sections, each containing data for a particular reaction, where the reaction type is designated by the symbol MT.
File 1 was for general information. File 2 contained information on resolved and unre- solved resonance parameters. File 3 contained information on smooth cross sections. File 4 contained information on secondary angular distributions. File 5 contained information on secondary energy distributions. File 6 contained information on secondary energy-angle distributions. File 7 contained information on the thermal neutron scattering law.