Three drawings are presented here with differing amounts of editing.Explanatory note follows diagrams.

May/June 1978 Newsletter was the first to publish the network topology. Minor edits have been performed.
1] The "*MASSEY" names were originally "*PRIVATE".
2] Node 54 was originally "3705B TORONTO".
3] Nodes 4 & 32 originally had "ENGLAND" as the second line rather than the circuit number.
4] Node 25 was originally below node 12. Nodes 56 and 51 were originally left of node 50. Node 47 was left of node 18 rather than below. Nodes 26, 28 and 45 were originally to the right and below nodes 37 16 and 45. (The changes have made the diagram smaller).

There is still only one loop in the diagram. The nodenames have been heavily edited.
1] Citynames were originally qualified in a rather pedantic style. I have preserved this style in the August 1981 diagram.
2] I have attempted to identify the customers who leased Alphas to serve their terminal population or inhouse system. Private nodes were more common in Europe where users were often charged for local phone calls made to an IPSA Alpha. This changed the economics.
3] Two Telex Interface machines are included in the diagram. They were not true nodes as their high order link to the network was used only for reloading the software.

August 1981 is the final diagram. I have tried to avoid editing nodenames. Some endnodes on the edges have been relocated to reduce the diagram width by two rectangles. The colouring scheme (legend on left side) reflects geographical location rather than node use.

These diagrams were produced by an APL program which had no knowledge of actual geography. It read a matrix from the network control file which gave the pair-wise connectivity of the network (e.g. node 1 Toronto connects to node 4 London, UK). This matrix was also used for less frivolous purposes such as calculation of route tables, identification of downline load target, etc.

Originally the only output device was an APL 2741 or ASCII terminal using the characters "_ | / \" to draw the rectangles and links. This form included high order link numbers within the rectangle. When the diagram became wider than 120 characters, the software was modified to print multiple vertical strips which could be trimmed and taped together to produce a wall poster.

Robert Bernecky wrote a supplementary program which took the 1x1 matrix of node numbers from the primary program and generated IPSA standard graphic matrices which could be displayed on a variety of vector oriented devices. Jean Gershater who edited the IPSA newsletter used this software and a HP7221 plotter to produce a topology diagram for the bimonthly newsletter.

The list of node names used within the Communications Department was not intended for public consumption. Names of private nodes included the customer name in some format such as "R+C" for Reckitt & Coleman. The internal node name often included a word describing the node's purpose such as "TEST" or "X25". During the period when the London office was moving from 118 Piccadilly to 132 Buckingham Palace Road, nodes were required in both locations. The internal name for node 32 was "Buck House". Fl. Lt. G H Robinson RAF (Ret.) sent an exaggerated letter of protest to head office. The letter made the tongue-in-cheek argument that the phrase was an insult to the Queen. This letter may have incited the creation of an altenate name list.

When the IPSA Newsletter began publishing diagrams of the network topology, an alternate list of node names was used. This has led to a small loss of information as the names of some private node customers have been forgotten. The distinction between 3705s and Alphas was eventually lost in the alternate nodenames.

Advances in network routing eventually killed the topology diagram program. The unit box system allowed eight links to be drawn from a box. Several nodes near the network centre used six links. Software changes completed in spring 1981 made the network quite tolerant of loops. The diagram dated August 1981 was drawn by manual placement of several loop member nodes.

When head office moved to the Exchange Tower in December 1981, the network acquired several loops within the Toronto computer room. Most U.S. and Western Canadian cities were provided with two connections to the network. The traditional scheme of a standard box with up to eight links of length one or two failed. There was some thought about changing the paradigm by fragmenting the drawing subnetworks. Perhaps representing some nodes with regions composed of two, three or four standard boxes would have allowed drawing a complete network with the restriction of four link orientations.