Sec. iv] THROUGH TALASH AND DIR 23
elaborate decorative friezes carved with diapers of the ` beehive ', and Amalaka ornaments reminiscent of the ruined Hindu shrines of Ketas, Malot, Amb, &c., in the Salt Range.
It is to these shrines also, and the temples I have since been able to survey at the two Kâfirkot sites on the Indus, that the Gumbat ruin shows closest resemblance in regard to all structural features. This will be evident from a glance at the plans and photographs furnished for the Salt Range ruins in General Cunningham's Reports and for those of the Indus sites in my own.4 The arrangement of the cella with the vaulted chamber above it, and the method of approach to the latter, correspond exactly to the dispositions observed in the main Ketas temple and in the pendant shrines B, C of the Bilot Kâfirkot. The latter approach the Gumbat ruin very closely in dimensions, and their resemblance in ground-plan would be still more striking if at Gumbat the porch had not suffered so severely from vandal hands. The destruction here of all architectural ornament has deprived us of the chance of proving in detail that the decorative motifs observed in the Salt Range and by the Indus were mainly derived from the later development of Graeco-Buddhist art in Gandhara, as I have suggested elsewhere.5
But the survival in the Gumbat porch of remains of the trefoil arch furnishes by itself a very characteristic indication. This architectural feature was long considered peculiar to the style of the old Kashmir temples, where it first attracted attention. But its presence is obvious in the far older remains of Gandhara Viharas and their sculptural representations, and M. Foucher, in his masterly analysis of architectural art in Gandhara, has proved that its true origin must be looked for there.G It is the prevalence of the trefoil arch in the Salt Range temples and those of Kâfirkot, which mainly accounts for the theory expressed by General Cunningham that their style was directly developed under Kashmir influence. The critical analysis of the historical records of the Kashmir kingdom has proved that its political power, which was supposed to account for this influence, was at all times restricted to a far more modest area than earlier writers assumed. It is only the rarity of architectural remains of later date in Gandhara, which has hitherto obscured the fact that the characteristics of the Salt Range temples of the centuries preceding the Muhammadan conquest can be traced to the direct development of that Graeco-Buddhist style, which had found its earliest and best known expression in the ruined shrines of Gandhara. Hence the special significance of the Gumbat ruin : it furnishes an example of this later development on ground which in art and culture was most closely bound up with Gandhara. There are no means of fixing the date of the temple with any approach to exactness. But taking into account what is known of architecturally related remains elsewhere, I am inclined to take the seventh and ninth centuries as the approximate limits of time.
The two long marches which carried me from Sado along the Panjkora to Dir were far too
rapid to permit of any close observation or inquiries. Nevertheless, I was struck by the absence on the hill-sides I was skirting of those large groups of ancient dwellings and towers which are so conspicuous on the spurs overlooking the Lower Swat Valley and Tâlâsh. Yet lower down by the river the fortified villages of the modern Pathan inhabitants were abundant. In the large and fertile village tract of Dir, beyond the Panjkora, it is true, I had to spend two days of enforced halt (May I and 2). But the conditions in which these were passed, as described in my personal narrative,? effectively precluded anything more than inquiries, and these did not reveal the existence
4 Cf. Archaeological Survey Reports, v., Pl. XXVI, XXVII ; Stein, Archaeological Survey, N. W. F. P., 1904-5, pp. 14 sq. (Figs. v, vi) ; Archaeological Survey, Frontier Circle, 1911-12, pp. 7 sqq.
6 See my Archaeological Survey, Frontier Circle, 191 I-12, p. 8.
6 See L' Art du Gandhdra, i. pp. 129 sqq., 139 sqq. ' See Desert Cathay, i. pp. 19 sq.