What rings are for
Figure 1.
This bit is based on, and extended from, an idea that arose in the letters pages of 'Spaceflight', the British Interplanetary Society magazine, in 1981.
Consider the problem of maintaining or repairing the outside of a rotating colony. As with any body of negligible gravity, a spacecraft could approach to within short distance of a rotating colony and maintain its position relative to the overall colony with little or no further expenditure of energy. The colony outer surface would then be moving past the stationary spacecraft as though it were a landscape that the craft were traveling over at speed. It would be difficult to perform any maintenance from this position other than to perhaps spray materials onto the habitat.
If you were rotating with the colony, working on the outer surface would be like working under a ceiling, but there would be no floor on which to build scaffolding. Even with a profusion of built-in hand-holds or rope attachment points a maintenance or repair team would be like mountaineers under a perpetual overhang.
The general idea is to make a mobile scaffolding consisting of a close fitting torus extending around the colony (colonies of cylindrical or axially-extended toroidal design are meant here) like a ring on a finger (Figure 1.) The torus cross section would be sufficient for men, entering via air-locks on the colony floor, to work inside. Its inner surface, that contiguous with the colony outer surface, could be open or have removable panels or doors to allow access to the colony. Work could either be carried out in vacuo in spacesuits or, if suitable seals could be extended from the torus to the colony surface, in a pressurized shirt-sleeve environment (Figure 2.)
Figure 2.
The entire torus would bear on the colony surface through sets of sprung wheels, possibly with automobile-like pneumatic tyres (Figures 3 and 4).
Figure 3.
Figure 4.
Some or all of the wheels could be driven by electric motors, powered by storage cells, when torus movement was required and they could all be steered together through up to 90 degrees to provide movement either around the colony (Figure 5.), along its axis of rotation (Figure 6.) or along some helical path having components of both the former possibilities (Figure 7.).
Figure 5.
Figure 6. 
Figure 7. 
Any point on the torus could be brought into close proximity with any point on the colony outer surface to allow maintenance checks to be made or repairs performed. The provision of torus movement around the colony is strictly unnecessary since a man could move around the torus to reach a desired point. On large colonies, however, the distances involved could be measured in miles and it may be necessary to move heavy items of equipment or materials quickly (e.g. serious leak repair).
For a maintenance ring to move around the colony cylinder it would be necessary to avoid obstructions, such as O'Neill's (O'Neill, G.K., The High Frontier, Jonathan Cape, London, 1977) external 'underground' trains or his mirrors attached to the cylinder by many cables.
Aside from inspection and maintenance, a further possibility of such a torus would be that if it were driven in the opposite direction to that of the colony rotation it could become stationary with respect to the surrounding space. It could then be docked with by spacecraft carrying materials (e.g. for colony repair). This would avoid all materials for repair having to be into the colony through axial docking point and then out to the maintenance torus through airlocks. After transfer of materials the torus could be spun up to colony speed again by ceasing to drive the bearing wheels and by applying gentle braking.
The window ceilings of the axially-extended toroidal colonies described by Hess (Hess, F. D., Demeter: Island in Space, Spaceflight, 21, pp. 393-396) and Hassall (Hassall, T., Towards Cities in Space, Spaceflight, 21, pp. 504-511) would provide another opportunity for the use of maintenance tori (Figure 8). For such an internal application the torus would not
Figure 8. 
any provision for pressurization. It could be entered via boarding lifts or stairs situated in the end caps of the colony. As with the outside of the colony mentioned above, the ceiling would need to be relatively free of obstructions to the passage of the torus. Rain pipes or sprinkler heads would need to be recessed in the ceiling.
The maintenance torus itself (or tori themselves) would need to be maintained, in a small version of the problem that made them necessary (or at least desirable) in the first instance. This problem could be solved by means of platforms attached to the colony surface but extending beneath the torus. These platforms would be retractable into the colony so that when flush the torus would pass over them without problems (Figure 9). The torus would be able to move 'sideways' into the platform and
Figure 9. 
could then be rotated around the colony so that all outer (or under) surfaces could be inspected and, if necessary, repaired. This would probably require working in full spacesuits.