The format of information recorded on magnetic tape, allowing a system to recognize, control, and verify the data. Tape format is defined at two levels.
The high-level format defines the data as it is presented to the magnetic tape subsystem by the host computer. This data stream consists of user data and labels, divided into sections (usually corresponding to files) by tape marks. The tape subsystem need not distinguish between user data and labels, but it recognizes tape marks. The data at this level may be divided into blocks, usually of equal length—typically a few kilobytes. This is done for physical convenience and has no logical significance. In some tape subsystems the data is continuous at this level, and division into blocks is done within the subsystem.
The low-level format defines what is actually recorded, as a pattern of reversals of magnetism, on the tape. The subsystem divides the data into blocks if this is not already done, and adds its own control information to each block and also at each end of the tape and each file mark. All formats include a degree of redundancy so that errors can be detected, and usually corrected, without reference to the host computer. In many formats an interblock gap (an area with no reversals of magnetization, typically half an inch long) is inserted between blocks so that the tape can, if necessary, be stopped and restarted between one block and the next (but see streaming). Where different formats are permitted on the same type of reel or cartridge, each format may include an identity burst or other means at the beginning of the tape to allow the subsystem to recognize the format.
Open-reel tape has been widely used for data interchange, so there are only a few accepted formats; all use ½ inch tape with nine tracks recorded in parallel. NRZ (nonreturn to zero) has a density of 800 bits per inch (bpi) which, with typical block lengths, allows about 20 megabytes of user data on a standard 2400 foot tape; it was introduced in the 1950s and is now virtually obsolete. PE (phase encoded), introduced in the 1960s, doubles this density and capacity. GCR (group code recording), introduced in the 1970s, uses a format in which data is recoded to give more powerful error correction, and packs 6250 data bits to the inch or typically 140 megabytes on a 2400 foot tape. All these formats are covered by ISO standards.
Cartridge tape was introduced after open-reel tape; it is used more for backup than for interchange, so standards are less essential. There is therefore a much wider range of formats. Some are defined as ISO standards, but others remain proprietary. Capacities extend to tens of gigabytes per cartridge.