THE PROJECT WAS PART OF AN OVERALL ATTEMPT TO COMPUTERIZE THE FACIAL IDENTIFICATION PROCESS. THE BASIC CONCEPT OF THE SYSTEM IS TO HAVE A COMPUTER SEARCH THROUGH MUG SHOTS AND MATCH THEM TO DESCRIPTIONS PROVIDED BY WITNESSES. THESE COMPUTER MATCHES ARE THEN GIVEN TO THE WITNESS FOR IDENTIFICATION. A SERIES OF THREE STUDIES AS CONDUCTED TO EXPLORE THE METHODS WHICH PERSONS USE TO RECONSTRUCT FACIAL IMAGES WITH THE HELP OF EITHER A SKETCH ARTIST OR AN IDENTI-KIT TECHNICIAN. THE THREE STUDIES DEALT WITH THREE TARGET POPULATIONS: WHITE MALES, BLACK MALES, AND WHITE FEMALES. IN ALL STUDIES THE WITNESSES VIEWED THE TARGET PERSON, THEN THE VERBAL INTERACTION BETWEEN THE WITNESS AND ARTIST OR IDENTI-KIT TECHNICIAN WAS TAPE-RECORDED. THE DICTIONARIES WERE DEVELOPED FROM AN ANALYSIS OF THESE RECORDINGS. THE DATA BASE INCLUDES 125 DESCRIPTIONS OF WHITE MALES, 40 FOR BLACK MALES, AND 42 FOR WHITE FEMALES. EACH OF THE DICTIONARIES IS PRESENTED IN TWO FORMATS. FIRST, FOR EACH OF THE 23 FEATURE TYPES (E.G., HAIR, CHIN, JAWBONE), THE VARIOUS DESCRIPTORS ARE PRESENTED ALPHABETICALLY ALONG WITH THE CORRESPONDING FREQUENCY OF OCCURRENCE. THE SECOND FORMAT PRESENTS THE DESCRIPTORS ALPHABETICALLY ALONG WITH FEATURES TO WHICH THEY WERE APPLIED AND THE FREQUENCY OF THEIR OCCURRENCE. THE REPORT IS REGARDED AS A DATA BASE REPORT, ESSENTIALLY FOR ARCHIVAL PURPOSES. FOR RELATED DOCUMENTS, SEE NCJ 47585-47588 AND 47590-47592. (AUTHOR ABSTRACT MODIFIED--GLR)