The Origin of the Triglyph Frieze

Abstract

SINCE the days of Vitruvius at least it has been the belief of most of those who have given attention to the matter that the triglyph frieze in Doric architecture arose from construction in wood. On this assumption our histories of architecture well nigh universally maintain that the triglyph represents the sheathed beam end of such primitive buildings, and that the metopes represent filling of what were originally openings through the frieze. I have elsewhere' sought to show how insecure is our ground for the belief in open friezes and I now presume to question the accuracy of the belief that the triglyph represents the original beam end. The source and the chief sponsor of the prevailing theory is Vitruvius, who sets forth his argument for the identification of the triglyphs as beam ends in his De Architectura, IV, 2, 4: non enim quemadmodum nonnulli errantes dixerunt fenestrarum imagines esse triglyphos, ita potest esse, quod in angulis contraque tetrantes columnarum triglyphi constituuntur, quibus in locis omnino non patitur res fenestras fieri. dissolvuntur enim angulorum in aedificiis iuncturae, si in Is fenestrarum fuerint lumina relicta. etiamque ubi nunc triglyphi constituuntur, si ibi luminum spatia fuisse iudicabuntur, isdem rationibus denticuli in ionicis fenestrarum occupavisse loca videbuntur. utraque enim, et inter denticulos et inter triglyphos quae sunt intervalla, metopae nominantur. oras enim Graeci tignorum cubicula et asserum appellant, uti nostri ea cava columbaria. ita quod inter duas opas est intertignium, id eroin~ est apud eos nominata. [Krohn.] This passage is usually accepted with implicit confidence as the final word on the subject. No one, so far as I am aware, has ever seen fit to inquire into the possibility of Vitruvius having chosen the wrong side of the argument, even though it is perfectly clear that he is here debating a matter in which he found some real difference of opinion among his sources. Let us therefore order a review of the case and see if errors of fact and of method do not