Magnetic core memory consists of an arrangement of a small piece of magnetic material (the core ) and current carrying conductors. The shape of the core may vary, cores took the form of wires, tapes or thin films, but toroidal cores seem to be most common. Cores usually consisted of a mixture of iron oxide or ferrite particles and a binding agent. The mixture is shaped and fired in an oven. Then individual cores are strung (usually by hand!) on fine insulated wires to form an array.
Each core would have two,three or four wires passing through it. The magnetic polarisation of the core (clockwise or counterclockwise around the wire) is used to store data. To write data to the core a current is applied to each of the wires passing through that particular core. The combined current of the wires passing through the core would be sufficient to change the magnetisation of the core. Of course each of the wires passing through the core would also pass through other cores in the array. However, the current from a single wire would be insufficient to change the polarisation of a core. To read from a core, pulses of electricity would be passed through the wires through it. The pulses would be of sufficient combined strength to change the polarisation of the core. It could be detected if the polarisation of the core was changed by these pulses. Obviously, this altered the information on the core in every second read operation, so after every read, if the state of the core had been changed, it would be re-written. The small size (each core was about the size of a typewriter 'o'), speed (average read/write time was about 1 microsecond) of core memory made it the most popular form of memory used in first and second generation computers..
Below is another example of magnetic core memory, but this time the individual cores are rod shaped. This was a much less common form of core memory, most likely because it was much more labour intensive to manufacture.
Here, individual arrays of cores have been placed on top of each other like the pages of a book to form a memory stack. Stacks usually comtained between 8 and 72 arrays. The stack, together with its associated read/write electronics forms a core bank or module of memory. The memory of a computer would be mad eof several such memory banks.
The IBM System 360 range of cmputers, released in 1964, represented two major changes in the way computers were manufactured. The first was the processing of data in 8 bit segments. This was also the first time that 8 bits of data was referred to as a byte ( the name can apparently be attributed to an employee of IBM at the time named Werner Buchholz. More important was the fact that this was the first large family of computers to use interchangable software and peripherals.
Pictured below is an example of the flexible printed-circuit "strips" of an IBM "Transformer Read-Only Storage" unit (also known as TROS). This is actually a type of random-access read-only memory of the type used in a number of machines in the System/360 family, and was used for microprogram code.
This is the type of read-only memory technology that preceded the advent of ROM chips. The TROS units were used in the 360 Model 40 CPU, the 2841 direct-access-drive controller, and a number of other I/O control units. The TROS technology operated on the principle of electromagnetic induction. To read out a word from TROS, a pulse was sent through a pair of leads on one of the printed-circuit strips. The pulse was directed either clockwise or counterclockwise around a square hole in the strip, depending on small holes that were punched in the strip. The current induced a magnetic pulse into a square rod that passed through the holes (one for each bit position), and the polarity of the pulse determined whether it read out a 0 or 1, as sensed by a low-inductance winding at one end of the rod.
Other machines, such as the 360 models 30, 50, and 65 CPU's, used ROM devices based on capacitance, rather than inductance. These were called "CROS" units (Capacitor Read-Only Storage).
At that time, it was IBM's policy to refer to "memory" devices as "storage", avoiding use of the anthropomorphic word "memory." IBM was attempting to disabuse people of the notion that computers could think, and so preferred not to use terminology that suggested that they could. At that point in history, very few people had direct contact with computers, so notions of "electronic brains" were commonly seen in popular culture of the time, and many lay people were afraid of what computers might be able to do.
The information about the TROS was kindly supplied by Russell P. Holsclaw.
