Mode of Construction. 469 meet, for the stones keep this direction only, close to the inner rim of the ring (Fig. 1S2). Some centimetres from this border, they begin to grow apart from each other, in such a fashion as to gradually form elongated triangles in the gap thus left, wherein large pebbles or quoins have been shoved in, whose function is to prevent displacement of the squares towards the horizon line. Whether greater solidity had not been assured to the dome with differently -shaped squares is beside the question. We have called attention to this arrangement to show that the method of the Mycenian architect, in what may be considered his masterpiece, is fraught with uncertainty and extreme rudeness. The work shows no advance upon the older Cycloprean con- struction. Thus, the mason filled the voids left in his otherwise Kir,. 182. — Mycente. Plan of last course of ihe Treasury of Atreus. fairly well-constructed and imposing domes with stones obtained by cleavage, exactly as his oldest predecessors, the Cyclopes, had done when they wished to stop the holes everywhere manifest between the enormous blocks forming the Tirynthian rampart, which had been torn from the hill-side ; some being quite rough, others barely touched with the tool. Regularity of masonry, then, is no more than the perfect unfolding of the Cyclopaean system : both incontestably belong to the Mycenian period. But we are less certain with regard to polygonal construction. At first sight it looks much more like the first system than does the second, with horizontal beds, (Fig. 93), but the resemblance is misleading. The materials beheld in Cyclopa^an work are manifestly irregular, whereas we feel that the irregularity of the third method is but on the