Valdemar Poulsen (1869-1942) received the earliest patent for magnetic recording in 1898. The son of a Danish judge, he attended medical school, but eventually became an employee of the Copenhagen Telephone Company at age 24.
Perhaps building upon the published reports of Oberlin Smith, he began experimenting with recording electrical signals on steel wire. On December 1, 1898, he filed a patent with the Danish Patent Office for the Telegraphone, incorporating a "Method of, and apparatus for, effecting the storing up of speech or signals by magnetically influencing magnetisable bodies".
Patent 8961 filed in the UK the following year clearly foresaw the development of today's computer tapes and disks: "Instead of a cylinder with a helical steel wire there may be uses as a receiving device a steel band, supported if necessary on an insulating material and brought under the action of an electromagnet. Such an arrangement has the advantage that a steel band of an desired length may be used. Instead of a cylinder there may be used a disk of magnetisable material over which the electromagnet may be conducted spirally; or a sheet or strip of some insulating material such as paper may be cover with a magnetisable metallic dust and may be used as the magnetisable surface. With the aid of such a strip which may be folded, a message received at any place provided with the new apparatus may be sent to another place where it may be repeated by passing the strip through the apparatus at that place." A working model of the Telegraphone was demonstrated at the Paris Exposition of 1900. A few recordings made at this event are still in existence.
Poulsen saw his new invention come to fruition in 1903 with the formation of The American Telegraph Company. The company manufactured the Telegraphones in the form of reel to reel wire recorders for use as telephone answering machines and dictating machines. Though the company ultimately failed, Poulsen went on to design a highly successful spark transmitter and other inventions, eventually winning honors and medals from institutions in his native Denmark and other countries. He died in 1942.
The wire recorder was the first practical application of Valdemar Poulsen's magnetic recording patent. Wire recorders were generally used for audio applications rather than digital data storage. They were rapidly phased out when inexpensive plastic-based magnetic recording tape became available. Plastic tape could be easily spliced and edited, was easier to handle, less wearing on recording heads, held more recording time on a reel and was less expensive.
The Williams tube (co-invented by Frederick Williams and Tom Kilburn) was used as fast access computer memory from the late 1940s until displaced by core memory in the early 1950s. It used a modified cathode ray tube to store electronic charge on a screen and was able to read the state of the charge as long as power was on and the state of the charged location could be refreshed. As such, the Williams tube is a precursor of the dynamically refreshed semiconductor RAM chips used in modern computers. It was the first RAM device available to store relatively large amounts of data. Its 2 to 8 kilobit capacity was considered quite large for its time.
Core memory, perfected by MIT's Jay Forrester (though with contributions from An Wang and several others), was the first reliable, random-access memory available for computer memory. It remained the dominant form of computer memory until it was supplanted by semiconductor RAM (specifically, the Intel 1103 dynamic random access memory (DRAM), introduced in 1970).
Core memory works by magnetizing small ferrite (powdered iron) doughnuts in either a clockwise or counter-clockwise direction. These two states corresponded to a digital '1' or '0.' Core memory is 'non-volatile,' meaning it retains its contents even with power removed. For this and other technical reasons, it is still used on some spacecraft today.
The round objects are the magnetic ferrite cores. Data is stored by setting the circular magnetic field in the core clockwise or counterclockwise to represent a binary "0" or "1". Core memory, invented in 1949, was the standard form of fast random access memory (RAM) used in mainframes and minicomputers for many years. It was first used in the 1951 MIT 'Whirlwind' computer and enjoyed a 15-20 year life span thereafter. While reliable and able to retain data after power removal, high assembly costs, power and space requirements left core memory unable to compete well against semiconductor memory as it developed and matured.
Plated wire memory
Plated wire memory, developed in 1957 at Bell Laboratories, uses the magnetic field produced by a flowing current in a wire to set or reset a magnetic field at the junction of intersecting wires. Like core memory, it is non-volatile and not sensitive to damage from radiation, which is a useful property in defense and aerospace applications. It has the advantage of being machine fabricated, rather than hand assembled.
Plated wire memory has been used in the Hubble space telescope, the Viking Mars lander and other satellite and deep space projects. It has also found a role in control systems for nuclear reactors. The memory module shown here was manufactured by Honeywell in 1964.
Delay line memory
A delay line memory carries a pulse stream that represents the 1s and 0s of computer data. The pulse stream is recirculated as often as necessary, forming a type of computer memory. Many different types of delay line have been built, ranging from tanks of mercury to radio waves, all working on the same principle of delay and recirculation. It is a predecessor of the DRAM chips now used in computers, which recirculate bits through a loop of memory cells. Today's chips can hold millions of characters - up to 32 million in recent designs
The sonic delay line memory exhibited, manufactured for use in a 1960s era Olympia calculator, stored 24 characters in a recirculating delay line. The bit stream, written at one end of the looped wire by creating shock waves that traveled down the wire, was detected at the other end a fraction of a second later and rewritten until the data changed or power was removed. .
For many years the mainstay mass storage device for computers (as well as audio and video equipment), the visual charisma of the twitching and spinning reels defined the public notion of a computer. Data was recorded in seven, then nine parallel tracks. Typical capacity was about 150 megabytes per reel for a 10" diameter reel of tape, but smaller form factors were also used.
Tape drives are effective for recording and retrieving large amounts of data at low cost, but are poor performers if the data to be retrieved is scattered throughout the tape because of the long time required to position the tape to the spot where the desired data is recorded. As a result, they were rapidly displaced by random access capable rotating magnetic storage (disk drives) in the 1960s.
The first tape recorder
In 1928, German engineer Fritz Pfleumer, demonstrated a magnetic recorder of his own design which used paper tape coated with steel dust. The inventor was granted a German patent for his invention. In 1936 the German National Court declared Pfleumer's patent invalid and previously described by Poulsen's original patents of 1898 and 1899.
However, in 1932 AEG, a German company, had begun the design and manufacture of the first magnetic tape recorder, naming it the Magnetophon. Another company, later to become known as BASF, produced the tape.
The sound quality of the Magnetophon was superior to that of other recording methods of the time, providing both superior fidelity and exceptionally low noise. The Germans put this quality to good use during WWII, recording and rebroadcasting the speeches of Hitler and confusing Allied analysts by making it appear that Hitler was in multiple locations nearly simultaneously.
After WWII, captured Magnetophons were brought to the U.S, where the technology formed the basis for the first products of Ampex Corporation.
The first magnetic tape drives
In 1951, the Remington Rand UNIVAC computer introduced the use of reel to reel magnetic tape as a program storage medium. (This machine used mercury delay lines for its internal RAM memory). The tape was unusual: Rather than using a plastic substrate, the tape was made of nickle-coated phospher bronze. It worked, but it was heavy. And the wear on the heads from the rapid passage of many feet of tape was unbelievable!
The first magnetic tape drive announced as a computer peripheral device for on-line storage was the IBM 726, introduced in 1952 and first shipped in 1953. It used a 7-track recording format across a half inch tape wound upon an 8 inch diameter reel containing 2400 feet of tape. The linear recording density was only 100 bits per inch, and a reel of tape contained 1 megabyte, or the equivalent of 12,500 80 character punch cards.
By comparison, a tape cartridge of today has a 384 track recording format and contains 100 gigabytes on 2000 feet of tape. Over the next several years, this capacity will expand eight-fold.
Reel of 1/2" 9-Track Tape (1970)
This reel of magnetic tape is typical of the type used in mainframe and minicomputer computer systems from about 1952 until the 1980s (although it is still in limited use today). Because magnetic tape was reliable and very inexpensive, it has remained a popular choice for medium to long-term data storage. One of its shortcomings, however, is that it is a sequential access storage medium, (in contrast to a random access one), meaning that one has to 'fast forward' or 'rewind' the tape until one reaches the data of interest - a time consuming process.
The TU56 DECtape Drive is a dual transport reel-to-reel tape drive. The DECtape drive served the purpose of a floppy disk drive on modern machines. Even though it was tape it was formatted into fixed size block and could randomly read and write them like the sectors on a floppy. This allowed the tape to have a normal file system on it. The main difference between the DECtape and a floppy is that the DECtape had a much longer seek time since the tape has to be sequentially read. The drive takes about 30 seconds to get from one end of the tape to the other. The DECtape is derived from the LINCtape on the LINC-8.
The operation of this drive is a little different than many other reel-reel tape drives. It doesn't have the capstans and pinch rollers to drive the tape and vacuum columns or other devices to control tape tension. It directly drives each reel with an AC induction motor. This method has some disadvantages: The drive can't stop fast enough to halt between tape blocks, so if the drive stops when reading sequential data, it must back up to reach the data it overshot.
Cassette and cartridge tapes
The reel to reel tapes used in mainframe computers were far too large and expensive to be used in minicomputers and personal computers. This led to the development and marketing of cassette and cartridge based tapes in a bewildering variety of formats to serve the small system market. Ironically, improvements in technology, improved handling and reliability, and favorable economics eventually resulted in tape cartridges that displaced reel tapes entirely for all classes of computers.
Despite annual forecasts of its imminent demise, the low cost, high capacity and removability of tape keep winning it markets in computer systems, especially in backup and archiving applications.
Magnetic drum storage
The magnetic drum is a rapidly rotating cylinder coated with magnetic recording material. Storage capacity is usually in the one thousand to fifty thousand byte range. Most drums had separate recording heads and electronics for each track of data, which made them expensive to produce. However, they offered relatively fast performance and were used where performance justified their high costs. They were displaced by magnetic disk drives as the latter became available.
English Electric Deuce Drum (1957)
This drum was part of the English Electric Deuce computer, a British computer. The rapidly-spinning cylinder (over 6000 rpm) has a magnetic coating that could store 8,192 words of 32-bits each on 256 tracks of 32 words each, or a total of 32 KB. This was considered a 'second generation' drum memory; first generation drums could be enormous and very heavy, yet their low cost made them a popular choice for mid-range computers in the mid to late 1950s. This drum weighs about 60 pounds. Note the handlebars for lifting. The cylindrical assemblies on top are for positioning the head stacks after the drum comes up to speed.
The first moving head disk drive
In 1955, commercial computers used tubes for logic, cores for short-term memory, head-per-track drums for intermediate storage and tapes on reels for longer-term memory. IBM had sent Reynold Johnson to San Jose, CA in 1952 with a direction to "produce something profitable". IBM's top electromechanical engineer for 18 years, "Rey" Johnson hand-picked a staff of 15 scientists, engineers and technicians. Working in a warehouse located at 99 Notre Dame Street, this new IBM team sought to develop a practical multiple disk storage system. It was to have the storage equivalent of 50,000 standard IBM punch cards and fetch the data in a single second.
Johnson's disk drive design was simple, but untried. The magnetic read/write sensors would have to be suspended only a few thousandths of an inch above a continuously rotating disk. If not accurately controlled, the magnetic pickups could strike the rotating platter. Although it was a high risk approach, Johnson accurately foresaw the benefits of future miniaturization of disk drives and pressed on.
In only twenty-four months, the team produced a functional prototype. It weighed one ton and occupied about 300 cubic feet of space. Fifty double-sided aluminum magnetic disks, each 24-inches in diameter, rotated at 1200 revolutions per minute on a common shaft, using externally pressurized air fed through the read/write heads to support the heads over a set of 24-inch disks. The drive's capacity was 5 megabytes, an unheard of accomplishment for its time. The head assembly was positioned by a hydraulic actuator.
The first such system, designated the IBM-350, was shipped in 1956 to Crown Zellerbach Corporation in San Francisco as an element of the IBM 305 RAMAC (Random Access Method of Accounting and Control). It can truly be said to be a critical milestone in the creation of the modern computer industry. This particular drive is now in the Smithsonian Institute collection.
In addition to his work on the RAMAC, Rey Johnson also distinguished himself in 1934 as the inventor of the mark-sense reader. The media for this device is familiar to the generations of students who took tests by filling in the space between the vertical dotted lines on the test firm, using the mandatory #2 pencil.
Burroughs ILLIAC IV 80MB Disk Drive (1972)
This large disk drive was part of the ILLIAC IV supercomputer, which used twelve of these disks for medium-speed data storage. Even though it was physically large, its performance was outstanding, even by today's standards, delivering some 500 million bits of information per second. It achieved this speed by having stationary read/write heads reading all tracks all the time, in contrast to today's disk drives that have a moving head seeking information as requested. The enormous (36" in diameter) disk platter could hold approximately 80MB (million bytes) of information. The disk itself is about a half inch thick. The disk was rotated by a belt driven from a motor.
This particular drive was actually used at NASA's Ames Research Center at Moffett Field.
Disk packs and cartridges
Both are magnetic disks that are removable form the recording/playback mechanism. The disk pack is a stack of disks on a common spindle. The cartridge is generally a single disk, though two disk cartridges are known. After the pack or cartridge was inserted in the disk drive, the recording heads were moved into place by a head positioning mechanism. Later versions had heads and disks located within a sealed mechanism to reduce the damaging effects of airborne pollutants and to improve the consistency of head positioning over the rotating data track on the disk. The original disk packs used 14" disks. Today's data cartridges use 3.5" or smaller disks.
IBM 1311 Disk pack drive (1963)
This is the first disk drive with a removable disk pack, and also the first using 14" diameter disks. Each pack was capable of holding 2 million characters. The concept of interchangeable disk packs lowered storage costs by making it possible to store information off-line yet also allowing that data to be made available by a quick and easy 'plug-in' procedure. Mainframe computer installations often had multiple 1311 drives, allowing rapid and reliable access to data, although the drive was initially introduced for use with IBM's smaller systems such as the 1401 and 1620.
The companion Model 1316 disk pack contained six disks in a protective plastic casing and weighed about 10 pounds. The disks rotated at 1,500 RPM. One disk pack could hold as much as 25,000 punch cards.
Diablo Disk Cartridge (1973)
This cartridge was based on the emerging disk drive standard developed by IBM, the Model 2315. It could store some 2.5MB of information while its single internal disk platter rotated at 1,500 RPM. It was removable, which allowed for convenient and rapid changing of data and programs as well as portability between computer installations. It was the "Zip" disk of its time.
Diskettes (floppy disks)
The first diskette was manufactured by IBM in 1969 for loading programs into the controller of a rigid disk drive. Publicly introduced in 1971, it was rapidly adopted as the mass storage device of choice for small business systems and then personal computers because of its relatively low cost and ability to remove and file the disk media. The disks, which have a magnetic coating placed on a thin plastic substrate, bend easily, hence the name "floppy disk".
First designed to use 8" diameter media, 5.25" media (1976) and then 3.5" media (1981) became predominant as improved technology increased storage capacity per unit area and form factors shrank. Various other sizes were offered, but were not adopted by the computer industry. The standards fights were brutal!
When first introduced, capacity was about 100 kilobytes (KB) per disk for the 8" diskette. Typical 5.25" diskettes offered 360 KB, then 1.2 or 1.6 megabytes (MB). The 3.5" units were typically .7 or 1.4 MB. Higher capacity variants were available, but never caught on because of expense and standards disputes. Smaller form factors had insufficient capacity to interest manufacturers.
While used mostly with computers, diskettes also stored data for digital cameras, sewing machines, and other industrial and consumer products.
Improvements in technology eventually produced very high capacity flexible disk drives and media with capacities exceeding 100 megabytes, but their higher costs precluded universal acceptance. However, they did become popular for information backup, off-line storage and data interchange between computers.
Eventually the Iomega "Zip" drive (100 and 250 MB) and the Matsuhita/Imation SuperDisk (120 MB) drives supplied the higher capacities users wanted. These achieved moderate success and are still available today from several sources.
Despite the limited storage capacity available, low cost and convenience created a long product life span for this class of product, which is only now beginning to be displaced by other types of disk drives.
The First Flexible Disk Drive:The IBM "Minnow"
The "Minnow" Flexible Disk Drive, first manufactured by IBM in 1970 (and shipped in 1971), was conceived and developed by David L. Noble and his team of IBM engineers and technicians. "Minnow" was the predecessor of a family of low-cost "floppy" drives which launched a major new segment of the computer industry. This commemorative object was given to Noble in 1978. The 8" floppy drive was initially developed not as a means of personal data storage but as a means of updating the 'microcode' of a large IBM disk drive system (the Model 3380). It was a read-only drive storing 81.6 KB of data.
In the early 1970s, magnetic cards were used to store information for word processing machines manufactured by IBM, Redactron, Xerox and others. The cards, the size of an IBM-style punch card, stored about a page of typed information. The manufacturing process was similar to that for magnetic tape: A magnetic film was coated upon a plastic substrate.
The magnetic card was displaced as word processing media by the advent of inexpensive flexible disk drives and media providing much higher storage capacity. However, the concept lives on in the magnetic stripe found on the back of today's credit cards, identification cards, and a variety of tickets dispensed by automatic fare machines. You may also find one on the back of your airline boarding pass the next time you fly.
Intel EPROM (1975-85)
The EPROM (Electrically Programmable Read Only Memory), is an integrated circuit that stores information permanently, yet is capable of being modified using a special 'programming' current. EPROMs are also 'non-volatile,' meaning they will retain their contents even if the power supplying them is removed. Before re-programming, the EPROM is erased by exposing the IC, via its quartz window, to ultraviolet light. This window is then covered after programming to prevent ultraviolet rays (from, for example, sunlight) from slowly erasing its contents. The printed circuit board shown here contains an example of both a protected and unprotected EPROM.
The role of EPROM has been captured by EEPROM (Electrically Erasable Programmable Read Only Memory) in recent years. We know it familiarly as "Flash Memory" because its EPROM ancestor was, literally, erased in a flash.
Optical Disk Drives
Optical disk drives use lasers to read and write on the disk surface. Attempts to design optical drives began in the 1960s, but the CD-ROM, which appeared in the 1980s, was the first successful optical storage device for computers. Originally a read-only device, read-write versions appeared in the early 1990s. The 500 to 700 MB capacity of CD-ROM made it a success as a distribution medium for software, which was undergoing explosive growth in memory requirements. And its 4.72" disk diameter meant it fit PC manufacturer requirements.
DVD drives, also laser read and written, were the next major success. With 4.7 GB per side, the disks were suitable for storing a digitized feature-length movie. But standards disputes and royalty issues delayed the introduction of writable DVD formats, and as of early 2002there is yet no universal format.
CD and DVD were created initially as read-only products for publishing data. Another class of products offering higher performance and inherent writability was also born in the 1980s. WORM (write once, read many) drives in 5.25" and 12" form factors appeared in the mid 1980s. Expensive and unstandardized, their appeal was limited. Rewritable magneto-optic media and drives in 5.25" and 3.5" form factors displaced the WORMs in most uses, but their high cost relative to pure magnetic storage kept their market penetration low. Still, the technical accomplishment was impressive.
This semiconductor memory is non-volatile - it remembers data even when power is removed. Applications include storage for digital cameras, music players, dictating machines and other portable devices. It is also used to store the "boot-up" data in computers and computer based telecommunications equipment.
Flash memory is usually packaged as removable cards that can be used to transfer data from a portable device to a card reader attached to a computer.
Several (non-compatible) formats are in use currently.
Specialized Alphabets and Symbol Sets
The Ubiquitous Bar Code
The packages of most modern products identify the contents both in words and with a code of black-and-white stripes. A laser scanner at the cash register of a retail store reads the price and other information from this bar code. The register records the price, adds up the bill, and tells a central computer how many items have been sold so that management knows when to reorder the item. Bar codes are also used in industrial settings to track factory lots or other objects, some as large as railroad cars.
By definition, a bar code consists of a series of parallel or concentric lines and spaces of varying widths that define short strings of alphanumeric characters. Bar code readers scan a light beam across the pattern of bars and spaces, detecting the reflected light pattern with a photodiode. The signal from the photodiode is then decoded into the appropriate symbol string for further processing.
Because the code is quickly read without human transcription of the data, accuracy and efficiency are high.
Bar codes got started in the late 1960s as an answer to the need for tracking two very different types of items: Supermarket products and railroad cars. However, the foundation of the technology can be traced to the late 1940s when Drexel Institute of Technology teacher Norman Woodland (right)and graduate student Bernard Silver began looking for a solution to the problem of tracking inventories. Eventually they took the idea to IBM, which sponsored further development. Woodland and Silver received a bar code patent in 1952. Their first design was a circular bulls-eye pattern which could be scanned in any direction, but the linear bars proved more practical for most applications.
In the early 1960s, David Collins of Sylvania devised a bar coding scheme to identify railroad cars. While different from the scheme used for packaged goods, the principle is the same. Sylvania's system was adopted by most railroads in the late 1960s. Collins started a new company, Computer Identics, which was successful in adapting bar codes for tracking industrial inventory.
Despite its advantages, adoption of bar coding was limited until the adoption of the Universal Product Code in 1973, which created a standard for use with retail packaging. George Laurer, an IBM engineer, is credited with the creation of the format, which was first used on packages of Wrigley's gum. Today, you can see UPC coding on the containers of most retail products. While UPC originated in the U.S., it has evolved to meet the needs of other regions and become a worldwide standard. Bar codes are now almost universally used for consumer and industrial goods in industrialized nations. A number of different coding formats (called symbologies) are used, depending upon the industry and geographic area involved. Eight formats are commonly encountered, but others exist.
Magnetic Ink Character Recognition
MICR to the acronym-loving, this technology was developed at SRI in the 1950s. It is in worldwide use for reading and processing of consumer and business checks. The familiar font, which can be read by both optical and magnetic scanning systems, is illustrated here: Written with magnetic ink on the bottom of checks and other documents, MICR characters allow the reading of the symbols by both machine and humans. They are used primarily in the financial industry.
This "juke box" is a type of robotic library, locating the selected information, positioning it in a reproducing mechanism and providing the playback (in this case, in an audio format). Its modern descendants include robot libraries for CD-ROMs, tape cartridges, video disks and DVD disks. They are found with computers, broadcasting equipment, home entertainment centers and automotive systems.
last updated: May 6, 2010
Original content: Copyright © 2000, 2001, 2002-2010 Museum of American Heritage
Trademarks are the property of their owners