Meg and Mercury


Work was started on a Mark 2 computer, often known as megacycle machine Meg, in 1951. It was essentially an updating of the Mark 1 concept, the design aim was to produce a faster more compact easier to maintain version of the Mark 1, which for user's convenience would include floating point arithmetic facilities. Apart from the incorporation of hardware floating point arithmetic, the instruction format and repertoire were similar to that of the Mark 1. Meg ran its first program in May 1954 and was perhaps the first floating point machine.

The changes made include:

Instructions were 20-bits long with the same single address format as the Mark 1.

The Meg prototype used a CRT as main store, with a provision to switch to magnetic core stores for the production version.

Use of semiconductor diodes helped to reduce the number of thermionic cathode by 57% compared to Mark 1. This along with other components produced a reduction in power consumed (12kw cf Mark 1's 25kw).Distributed electromagnetic delay lines provided a cheap, flexible implementation for many of Meg's internal registers. A single drum backing store was provided for the prototype.

The speed was increased by operating the cathode ray stores in parallel. Improved circuitry design helped increase speed and reduce the size of the basic machine.


The production Meg, was made in cooperation with Ferranti. It was known as the Ferranti Mercury. This only differed from the Meg in details of the order code repertoire and in the main store technology. Developments in storage systems meant that Mercury was equipped with a Magnetic core matrix to replace the cathode ray tube store used in the MEG. Mercury's main store was 1024*40 bits of ferrite core, at 10-microseconds cycle time for a 10-bit short word. It was divided into 4 blocks for addressing purposes, consistent with the usage of short (10-bit) medium (20-bit) and long (40-bit) words. Mercury had a backing store consisting 4 drums each holding 4096*40-bit words and having a 17.28 millisecond revolution time. Parity checking was provided on main and backing storage.

The first Mercury was delivered in August 1957. One was installed in the Manchester computing machine laboratory in Feb. 1958. Notable software development was the high-level language Mercury Autocode. Tom Kilburn said "Programming for the machine is very simple, using the Autocode technique. The Mercury Autocode system can be learned by a programmer in a few days and has made it possible for anyone to write his own programs."

Ferranti sold 19 Mercury machines. The Manchester Mercury was used until Jan. 1963.