Vertically interlocking load-spreading unit paving systems offer many economic advantages, among which are the reduction of sub-structure and diminished maintenance costs. The designs that have reached the market have often relied on complex, joinery type connection detailing and, because of this, have never fully achieved their possible economies. The patented G-Block System results from an explor-ation and analysis of solid geometries. Tne close-packing characteristics of tetrahedra have been developed into the G-Block range of blocks and slabs which, when laid, exhibit excellent load distrib-ution characteristics. The system includes an edge block, a reinstatement unit and also a sonhist-icated machine-laying technique. The search for a vertically interlocking paving block has been, for many people in the industry, like the search for the philosospher's stone -a discovery that would turn base metal into gold. Certainly, an effective vertical interlock would offer two immediate advantages over most current systems. First, in improving the lateral load-spreading characteristics of a paved area it would reduce the sub-structure requirement sign-ificantly and thus lower initial capital cost. Second, because of increased stability and resistance to 'punch-in' it would reduce main-tenance costs. There can be little doubt that this would not only affect existinG markets but must, in time, introduce completely new market areas to the concrete block industry. Current practice requires -in the broadest terms -an over-thick surface course on a sub-structure designed to cope with the worst poss-ible condition. It seemed to me that -theo-retically at least -there could be two routes for design rationalisation. First, the sub-structure could be improved to give total sup-port to a block which was substantially reduced in thickness and which was expected to perform simply as a biscuit-like surface. Second, the units or blocks could be designed to have posi-tive structural interdependence, thus allowing for a downgrading of substructure. The first of these options was judged practically unattain-able while the second seemed a direction for fruitful research. Many others have perceived the logic of design development of structural interlock on the vertical axis but, with the benefit of hindsiGht, it is possible to isolate an error of design thinking in previous examples. At a larger scale, in precast concrete buildinG for example, mechanical jointing systems are commonplace apd are usually descendants of traditional joinery techniques -tongue and groove, dovetail, mortice and tenon, etc. In my view, the small scale of most paving block systems precludes the efficient use of this kind of connection technique. The disaavantages of complex interlock can be listed : 1. Weight and size. Naturally, if a block is to have particular constructional detailing on its edges, it tends to become larger and heavier than a non-connectinG block. The economic repercussions are obvious. Not only are handlinG difficulties increased at the factory and on site but, in addition, the surface area of paving per truckload gets smaller. 2. Damage. The more precise the connection detail, the greater the risk of damage to it during handling. 3. Difficulty of installation. It has often been the case that a connection detail that is beautiful in theory demands, on site, the ki~d of care in installation which is either unavailable or expensive. For these reasons we rejected the 'ed~e conn-ection' approach and defined the problem in new, and fairly rigorous terms. We were looking for a block confiGuration which had no 'joinery' type connections, which was easy to manufacture, transport and lay and which would be unlikely to sustain accidental damage in handling. We were looking for a block which would transmit loads laterally and which would close-oack as a fund-amental characteristic of its three-dimensional geometry. It took some time. It was clear that all exist-ing systems were based, topologically, on cubic packin9 -modified and shaned or not -and we felt that further research in this area miGht stimulate slight improvements but was unlikely to produce the second Generation block we were looking for. The tetrahedron provided the key. The tetra-hedron is a solid with four surfaces, each an identical equilateral triangle. It is usually shown as a three-sided pyramid (Figure 1). When the tetrahedron sits on one of its triangular surfaces. as in Fioure 1. then all horizontal sections through it will be triangular. Figure 2 shows the tetrahedron posed on one of its six edges. In this position the top edge and the bottom edge are horizontal, and at right-angles to each other. A horizontal sect-ion taken at mid-heiaht throuah a tetrahedron in this position is a square -the 'equatorial square'. All other horizontal sections are rectangular.