The façade of the Clunisian abbey church of Saint-Gilles-du-Gard is one of the most complex and elaborate examples of the profound influence of antique architecture and architectural sculpture on the late Romanesque style in Southern France. Since the 19th century, several generations of scholars have studied this famous monument, trying to establish the history of its construction, explaining the stylistic diversity of its sculptural program and commenting on its architectural and artistic concept. One of the major aspects consistent with nearly all previous theories is that the design of the façade, considered as incoherent in its actual state, had been subjected to major changes during the building process. Attempts to reconstruct the original hypothetical design, by various authors such as Marcel Gouron, Victor Lassalle and Whitney S. Stoddard, were based on the tacit assumption that the Romanesque project was founded on a logical concept similar to those found in the later Renaissance and classical period. Idealistic visions of what the "original" façade would, could or should have looked like led either to a selective perception of authentic archaeological evidence, or to misleading interpretations of the architectural structure. Since the late 1970’s no major study of the façade of Saint-Gilles has added substantially new or convincing arguments to this futile discussion, from an archaeological point of view, and both Willibald Sauerländer and Dorothea Diemer, author of the latest architectural analysis of the monument, have noted the necessity of authentic fundamental research before resuming the debate. The evidence that no further progress could be achieved without a profound and systematic archaeological survey of the entire façade became the starting point of our thesis. From 1999 to 2002, three whole years were spent on a complete and exact stone-by stone scale drawing of the façade and its foundations, as well as of the aisle walls of the adjacent crypt and nave, consisting in all the exterior and interior elevations and horizontal and vertical cross-sections. The combination of measurements taken manually from a mobile scaffolding and a precise survey assisted by computer and laser technology (theodolite, laser level) resulted in a highly reliable type of drawing which benefits from both a thorough macroscopic examination of all the visible surfaces of the monument and a dense network of precise topographical threedimensional references. An archaeological survey alongside the foundation walls, undertaken in 2004, added substantial evidence to the results of the stone-by-stone analysis. The survey revealed that the Romanesque foundation was built upon earlier structures, some of which were either removed or incorporated into the central projection of the western wall of the crypt on which the façade rests. Resistant to the weight and pressure of the new structure, the remaining masonry caused severe damage to the ashlar masonry when parts of the façade started to move downwards during the building process. The deformations, hardly perceptible to the eye but measurable, must have become obvious almost immediately, as the masons tried to level the courses with thicker joints and with angles bent back into a vertical position, which soon began inclining again due to the slow but continuous movement of the façade. The huge crack in the middle of the foundation would actually have begun to open before work had started on the upper level of the façade itself. As a major threat to the stability of the planned structure, the crack seems to have been responsible for the first major change of the project which included a protruding protiro inspired by North Italian prototypes, the pedestals of which were never executed as such but were instead replaced by blocks of a considerably lesser size, and their lion figures put to a different use. By the time the structure had reached the level of the main lintel the protiro was definitely abandoned, the protruding pairs of columns were subsequently shortened and fitted with cornice blocks at a lower level, and the main blind arch on the inside was downsized to the same level as the secondary arches of the south and north portals. At that time the centre of the façade must have been left open, with the decision of building a nonprotruding archivolt being taken at a later date. The orphaned corbels for the protiro gable were finally inserted above the major frieze, and the façade was completed. Our analysis of the façade has established that the sculpture was entirely prepared in advance. The evidence of reused antique stone material, which the sculptors chose for their excellent quality, could explain some of the irregularities and the incoherent assembly of the superior registers of the façade; the size of the three lintel blocks being obviously insufficient to match the adjacent blocks of the figured frieze. On the other hand, no proof could be found for any of the major modifications suggested in the former studies of the façade, the sculptured blocks having been assembled in a regular process without obvious interruption, with the exception of the main archivolt, the corbels and the retro-fitted parts of the underlying cornice. Our study contradicts former opinion that the building of the façade was delayed after the western wall of the crypt which the façade is based upon had been completed. Both levels are, in fact, inseparable and closely related. Incompatible with this technical evidence is the early inscription in the south wall of the crypt, which commemorates the foundation of the church in 1116, and which should not necessarily refer to the including masonry. The same conclusion applies to the epitaphs in the western wall dated 1129 and 1142: the archaeological survey has proven in at least two cases the absence of any contemporary tomb underneath the funerary inscriptions, which could have been cenotaphs for older tombs removed for the foundations of the church. Comparisons with Saint-Trophime at Arles and other related monuments in Italy tend to confirm that the sculpture cannot be earlier than the late 1170s or early 1180s. The very similar crypt of Saint-Trophime, which was possibly completed for the translation of the relics of Saint Trophimus in 1153, should hint at a close date for the crypt of Saint-Gilles. The focus of our study of the façade of Saint-Gilles excludes the making of the sculpture itself, nor do we discuss the formerly and much debated problem of the artists, seriously questioned by Dorothea Diemer, Willibald Sauerländer and analyzed by Andreas Hartmann-Virnich in his studies on the sculpture of Saint- Trophime at Arles. Instead, our archaeological approach reveals in great detail the techniques, and the strategies which the builders developed and used in order to integrate the earlier structures, and in order to counterbalance the negative effects that resulted of their insufficient knowledge or carelessness during the building process. The history of the building of the façade of Saint-Gilles, told through the masonry itself, offers a fascinating insight into the world of the builders of one of the chief masterpieces of Romanesque architecture in the Mediterranean.