709,407. Digital electric calculating-apparatus. ECKERT-MAUCHLY COMPUTER CORPORATION. June 28, 1948 [June 26, 1947], No. 17135/48. Class 106 (1). [Also in Group XL(c)] A computing machine comprises electronic trigger units interconnected to form a plurality of computing devices each responsive to pulses representing quantitative values to perform a predetermined computational operation upon such values and to provide a terminal pulse signal upon the termination of each such operation, and electronic programming means arranged to be stimulated by particular ones of such terminal signals to programme the sequence of such computational operations at electronic speeds in any predetermined relationship, whereby the results of any of the operations may be used in other subsequent operations in any of the computing devices to form a programmed routine of operations. The machine is primarily intended for computing large families of solutions all based on the same programme of operations. The entire computing schedule is carried out automatically after the sequence which it is to follow has been set up manually. The various elements of the machine fall into three classes : (i) arithmetic elements, (ii) memory elements and (iii) control elements. The first class includes means for carrying out the four rules, taking square roots, and comparing magnitudes ; the second comprises " external " memory e.g. punched cards, perforated tape, &c., and " internal " memory e.g. switches, relays &c. in the machine the third includes a " master programmer " unit controlling programme controls on the other units. These latter controls, when initiated by a programme pulse, cause the unit on which they are located to operate in accordance with the settings of various switches on the unit. The machine described comprises a series of forty vertical panels of uniform height and width extending round three sides of a large room, Fig. 1. In addition there are three truck-mounted function tables for connecting in at alternative parts of the system, a record card reading apparatus for entry of data, and a card punch (referred to as a "printer") for automatically recording results. A power pack is located at a distance from the main equipment, and there is also a ventilating system to extract the hot air generated by some 18,000 valves. The faces of the panels comprise switches for determining certain functions of the circuits therein, and for interconnection of the panels in variable relation for the transmission of numerical data and programme pulses there are link cables and plugs and so-called "trays" in which conductors are spaced with shielding between to minimize reactance eflects &c. The circuits are built up from standardized. sub-units including flip-flops, Fig. 3 (not shown), binary rings, Fig. 4 (not shown), decade counters, Fig. 5 (not shown), and other counters, gates Fig. 6 (not shown), buffers, Fig. 9, (not shown), pulse standardizers, Figs. 12 (not shown), and 13, &c. (see Group XL(c). Initiating unit.-This unit (panel No. 1, Fig. 1) permits manual initiation of the operation of the machine by the formation of a starting pulse properly synchronized with the basic pulse times of the system to ensure positive starting without the liability of splitting a pulse. When a switch A75, Fig. 13, is closed, connecting the grids of valve A76 to zero potential, a condenser A77, connected to +75 volts, discharges through a 4.7K resistor. As the condenser, with a time constant of just over two milliseconds, discharges, both grids of the double triode drop in potential and the valve ceases to conduct, the consequent rise in anode potential causing valve A80 to conduct. The cross connections of valves A80, A81 cause A81 to be cut off, and the negative pulse at the anode of valve A80 is transmitted to a further circuit which responds only to negative pulses. Thus, the positive pulse which occurs when the switch A75 opens and the valves A76, A80, A81 return to their initial conditions has no effect. Moreover, the time constants for the discharging and charging of the condenser A77 are sufficiently long to prevent spurious operation due to chattering of the switch contacts or other transient effects from causing more than one negative pulse at the output of the circuit. The starting pulse, once communicated, causes the machine to store data supplied through the card reading apparatus, after which the machine automatically functions computatively on the stored material at an electronic speed. The initiating unit comprises also controls for turning the power on and off, for initial clearing and selective clearing of accumulators as well as programme controls for the card reader and printer. Testing devices are provided including D.C. and A.C. voltmeters and a D.C. voltage hum oscilloscope. Cycling unit.-This unit (panel No. 2, Fig. 1). produces the pulses (see Fig. 2) which are used to represent digits and to control the operations of the various units. It comprises a standard quartz crystal oscillator emitting 100 KC sine waves which are shaped by a pulse standardizer, similar in arrangement to' that described above, and fed into a 20-stage electronic counter. Pulses emitted from various stages of this counter are used to gate flip-flops, or in some cases are used directly as digit or control pulses, in a cycle based on a pulse time of 10 microseconds and repeated every 20 pulse times (i.e. every 200 Ás). Each cycle or frame of pulses, which is known as an " addition time," since it is the maximum time required for the addition of two ten-digit numbers, is signalled by a programme pulse (CPP) ; the 9P, the 1<SP>1</SP>P, the 1, 2, 2' and 4P pulses are commonly used to represent digits, the 1, 2, 2<SP>1</SP> and 4 pulses being used particularly where information stored in static form is converted into pulse form, e.g. in the high speed multiplier, the function table, the divider-square-rooter, and the constant transmitter. The 10P pulses are used only in the accumulators. They serve to cycle each counter around back to the position it starts from when a number and/or its complement is transmitted from an accumulator. The carry clear gate CCG, which lasts from pulse time 11 to pulse time 17, is used for clearing the accumulators, if desired, after transmission of their contents, and also to allow a carry pulse to pass from a decade counter to the higher denomination where appropriate. In delayed carry over the first reset pulse RP is " gated by the carry clear gate to reach the next denomination ; in direct carry over (i.e. carry over which results from carry over) the pulse which necessitates carry over is the one which the carry clear gate allows to pass to the next higher denomination. The second reset pulse RP (and in some cases the first RP pulse) resets carry over flip-flops which have been involved in carry over. The 1<SP>1</SP>P pulse is used in connection with the " fugitive one." A switch on the cycling unit enables the operator to suspend continuous operation, and the twenty steps of the addition cycle are then carried out each time a push button is depressed. Another position df the switch allows each individual step to be controlled by the push button. The forms of the pulses generated can be viewed on an oscilloscope built into the cycling unit. All the pulses are passed through transmitters for power amplification before passing into the several lines of the cycling trunk. Accumulators.-The accumulators, which occupy twenty of the panels, Fig. 1, serve both as memory and as arithmetic units, performing addition and subtraction. Each can store, and operate upon, a ten-digit, positive or negative number, while two accumulators can be interconnected so as to store or operate on a signed number of twenty digits. Repeat programme controls make it possible for an accumulator to receive or transmit a number repetitively from one to nine times when a particular control is stimulated, while cable shifters make it possible to receive an incoming number shifted to the right or left. Thus, by repeated addition, the accumulator can carry out multiplication of a number by a constant having one or more digits. There are five digit input channels for each accumulator, through any of which a ten-digit signed number may be received. The circuits of an accumulator comprise three groups. First, the programme control circuits comprising receiver and transceiver plug-in units and their associated switching circuits i.e. programming memory, and the repeater circuit. Second, the common programming circuits which directly control numerical circuits by opening certain gates and, in some cases, introducing certain pulses into the numerical circuits. These circuits determine which operation the unit shall perform. Third the numerical circuits comprising ten decade ring counters and an additional binary ring for registering the sign (plus-minus). The operation performed by the numerical circuits is determined primarily by the various switch settings of the particular control circuit used. The receiver circuits cause the accumulator to add to its present contents any ten-digit signed number arriving through one of five input channels SV ... SV5, Fig. 33C, under control of receiving gates N81. A positive programme pulse arriving at the input N6, Fig. 33B, of a receiver N4<SP>1</SP> causes a valve N52 to conduct and send a negative pulse to reverse an associated flip-flop N4. An inverter N53 cuts off causing a cathode follower N54 to conduct thus passing a positive gate voltage through deck 3 of the operation switch which. when set at one of the five receive positions x, #, γ, #, #, (say α) directs the positive gate voltage to SU1, through interconnector cable to ST1, causing a valve N82 to conduct and an inverter N83 to be cut off. The positive gate from N83 opens the gate tubes N81 of the alpha group SV, Fi