The Grande Hotel at Ouro Preto was a small building in a remote location. Part of the reason it assumed such importance was its presence in a remarkable exhibition held at the Museum of Modern Art in New York in 1943: Brazil Builds: Architecture New and Old, 1642–1942. Curated by the museum’s co-director Philip Goodwin, himself an architect, with photographs by G. E. Kidder-Smith, it was vital in reinforcing the idea of Brazil as a modern nation, and the architectural careers of Costa and Niemeyer in particular. Its impact in Brazil was considerable, too, aided by the production of the catalogue in a bilingual edition (the Portuguese title was Construcao Brasileira).

However, the emphasis on what might be termed the sculptural aspects of the building is made at the cost of the building’s practicality; in other words, the concern for surface effect overrides the expected Modernist preoccupations of light and space. Bruand wrote of the disastrous quality of the rooms: their ‘total’ discomfort, their claustrophobic form ‘like narrow corridors’, their scale ‘visibly sacrificed’ to the overall visual effect. Access to the upper floors is via a spiral staircase, which is ‘impractical and dangerous’ for the old and children, and robs each apartment of valuable living space.

At this point Costa became involved. Writing to Andrade from New York, where he was busy building the Brazil Pavilion for the 1939 World’s Fair, he politely expressed alarm that Leao’s design was a capitulation to neo-classicism, a style that Costa himself had recently abandoned, but which still had numerous influential adherents. Costa wondered if the project marked a ‘rejection’ of the Modernism with which he was himself now increasingly identified. He encouraged Andrade to commission a further study with Oscar Niemeyer as the architect; Andrade agreed.

These ideas of Costa and Freyre, with all their manifest contradictions, were played out with remarkable clarity in a handful of small buildings. Chief among these is the Grande Hotel in Ouro Preto, whose name belies its small scale. It was designed by Oscar Niemeyer in 1938–9, with the well-documented involvement of SPHAN, and it was completed in 1940. The government of the state of Minas Gerais, of which the town is the old capital, first looked at building a new hotel on 1938, to capitalize on the city’s touristic potential, and considered a number of designs.

Freyre’s affection for the casa grande was shared by many important Brazilian intellectuals of the 1930s, almost all of whom can be identified with the left. Among them was Costa, who included it among the types of buildings of the historical past that he wished to defend. Like Freyre, he moves into a wistful, nostalgic mode when describing the casa grande.

One of the crucial intellectual sources for this unusual attitude to the past was the sociologist Gilberto Freyre. As Cavalcanti notes, Freyre’s work was integral to the development of a concept of a modern Brazilian identity. His great idea was racial democracy: that is, Brazil as a racial democracy at a time of ubiquitous racism. Brazil’s race relations, he argued, were uniquely liberal, in spite of, or in some ways because of, its long history of slavery. The crucial work in Freyre’s oeuvre, and the best known outside Brazil, is Casa Grande e Senzala.

The argument is principally that architectural tradition is not invested in surfaces, but rather in traditions of building, wherever they may be found. He had not realized, he wrote later, that ‘the real tradition was right there, two steps away, with our contemporary master-builders . . . it is enough to make up all that lost time by extending a hand to the master- builders, always so scorned, to the old portuga of 1910 because, say what you like, it was he, alone, who was guarding tradition’. Benzaquen de Araujo uses Costa’s term ‘saude plastica’ to indicate an ideal relation between past and present. This concept, ‘plastic health’ loosely translated, indicates a way of thinking that describes a close, and essentially Modernist, relation between form and function; aesthetics cannot exist alone, but must be accompanied by an interest in the ‘primordial’ activity of construction.

But first we should look more closely at the intellectual context that makes possible this sophisticated and nuanced engagement with the past. One figure in particular stands out, Lucio Costa (1902–1998), born in Toulon (France) and educated in Newcastle, Montreux and finally Rio de Janeiro, where he graduated as an architect in 1924 from the Escola de Belas Artes. Costa soon established a partnership with Warchavchik, and in 1930, only six years after graduating, became the director of the Escola de Belas Artes. His reign there was controversial, and he was forced to resign after only a year.

The richness of early twentieth-century architecture in Brazil has been consistently devalued in favour of that of the colonial period. This is both a problem and a paradox. Many of the most admired structures of the colonial period were openly a means of maintaining a feudal slave society, yet their advocates were in most other respects politically progressive.

This unmistakable piece of Haussmannization was accompanied by some flamboyant public buildings built in Rio in the Second Empire style, including the Biblioteca Nacional (1910) and the Museo Nacional de Belas Artes (1908). However, the Teatro Municipal (Francisco de Oliveira Passos, 1909) stands out, an extraordinary confection of marble, onyx, bronze and mirrors, based on Charles Garnier’s Paris Opera, with all materials imported from Europe.

There is a history of pre-colonial indigenous building that is rarely part of any architectural discourse. And there were other European presences besides the Portuguese. In 1816 the Portuguese emperor Joao VI brought a number of significant French artists to Brazil, including the architect Grandjean de Montigny, who designed the first significant French-style building in Brazil, the Escola de Belas Artes in Rio de Janeiro – thus began a significant period of French influence.

However, as at the Architectural Review in England, SPHAN served up a highly idiosyncratic version of the past. SPHAN, much influenced by Costa, has always emphasized the architecture of the Portuguese colonial period, representing it as the one true historical architecture. Its principal sites of the Baroque, now protected under the aegis of IPHAN (Instituto do Patrimonio Historico e Artistico Nacional, the successor to SPHAN) or UNESCO, or both, include the colonial hutches of Rio de Janeiro; most of the former capital of Minas Gerais, Ouro Preto; the historic centre of Salvador da Bahia in the north-east; the eighteenth-century centre of Olinda in the state of Pernambuco; and the missionary towns of the far south-east of Brazil.

An early example of this tendency is the house that Gregori Warchav chik built for himself and his wife (1927–8, cited before). As much as this house broke with tradition, it also made explicit reference to the local context: its tiled roof, whitewashed facade and extensive veranda clearly refer to the vernacular architecture of the casa grande, of which more later. But Warchavchik is an isolated case. The crucial ideas in this context come from an institution, SPHAN (Sociedade do Patrimonio Historico e Artistico Nacional, or Society of National Historical and Artistic Heritage), created in 1937, and in which Lucio Costa was closely involved at the same time as designing the sensational MES. SPHAN was an organization that understood the codification and protection of the past as integral to the Modernist project.

As I described before, Brazil’s Modernism is polyvalent, plural and often frankly contradictory. Nowhere are these qualities more apparent than in the relationship of Modernism to the past. Brazil’s official view of itself is, to appropriate Stefan Zweig’s description, ‘the land of the future’. It has been the land of the future for the best part of a century, and this idea has become something like an article of faith.