The entry of administrative data includes information about a case, document, individual, and document part in the system files. The document-part image is acquired by scanning the entire document and selecting the desired part by using a 6.5 x 6.5 cm ruled square. Once the block of writing is selected, the document is rescanned, producing an enlargement of the part on the screen. By adjusting the contrast between the writing and the paper surface to conform to that of the original document, the examiner can present features, such as tapered beginning and ending strokes or light connecting strokes, as they appear on the original document. The examiner may also remove unwanted features from the document image. The feature extraction process can begin according to the type of writing indicated on the document-part data entry screen. The first step is the text-insensitive feature extraction (TIM), an automatic process in which FISH marks the exact center of the part and draws eight vectors out to the edges. The system then uses black-and-white pixel chain formations and auto-correlation functions to mathematically treat and store the writing. The TIM is then saved, and isolation for selected features begins. This involves circling the desired letter or combination. The next extraction deals with tracing or line-following. A set of measurements then marks the height of the base ovals. At least 10 measurements are taken. After the interactively measured features are recorded, a screen appears that lists the number of measurements taken, the median measurement, and the standard deviation for each category. This article also describes how to compare an entered writing against those previously recorded in the database. The article concludes with a discussion of how the U.S. Secret Service, the custodian of FISH, provides FISH-related assistance to outside agencies.