Why the Old Halls Still Win

AAsk a hundred musicians to name the finest rooms in which they have ever played and the same small group of nineteenth-century halls will appear on most of the lists. Nearly all are rectangular, narrow and small, built before acoustics existed as a science. Ask the same question about the halls completed since 1960, when the terraced or 'vineyard' plan — the audience seated in blocks around a central platform — began to spread, and the answers scatter. The discrepancy embarrasses a profession that now models rooms in software before a brick is laid, and three explanations of it circulate. The first locates the quality in the geometry of the room, the second in the listener rather than the room. The third denies that the old halls are as uniformly fine as the lists suggest, and asks what became of the ones nobody mentions.

BAnneke Vermeer, who directs the applied acoustics group at the Delft Institute of Building Physics, defends the first explanation with a single quantity. Sound that reaches a listener from the side within the first eighty milliseconds after the direct sound arrives is heard not as a repetition of the music but as width: the orchestra seems broader than the platform it sits on, and the listener feels surrounded. A narrow room supplies such reflections from its side walls; a wide one cannot, since the sound has too far to travel and arrives too late. In a survey published in 2016, Vermeer's group measured the lateral energy fraction in forty-one halls and set the figures against ratings collected from 1,730 concertgoers. Halls under twenty-five metres wide averaged 4.6 out of 6; those above thirty-five metres averaged 3.4. Vermeer states the limits plainly: her regression accounted for forty-six per cent of the variation in the ratings, and the narrow halls in her sample were also the oldest and the smallest, so the measurement cannot be disentangled from age and size.

CRafael Duarte, who runs the psychoacoustics laboratory at the University of Coimbra, holds that the argument has been conducted about rooms when it should have been about listeners. In 2019 his group recorded the same orchestral passage in twelve halls using a head-shaped microphone and played the recordings back to ninety-six listeners over headphones, so that nobody knew which hall was which. Under those conditions the rectangular halls still came top, but the gap narrowed sharply: the advantage they hold in the published rankings roughly halved once the room could not be seen. Ticket price and sightlines, he argues, are doing work credited to walls. Headphone reproduction is at its weakest exactly where halls differ most, in the two lowest octaves, and every judgement he has gathered was made in a laboratory rather than in a hall with an audience, a coat and an interval.

DEleanor Braithwaite, an architectural historian at the Marchmont Institute in Melbourne, thinks both camps are studying a sample that has already been sorted for them. Before steel trusses became ordinary in the 1890s, a timber roof could not comfortably span more than about twenty-two metres, so a hall of any capacity was necessarily long and narrow: the shape was a consequence of carpentry, not of listening. From municipal archives she has assembled records of 214 halls built in Europe between 1820 and 1910, and finds that only thirty-one remain in concert use. The rest were converted, demolished or rebuilt, and the survivors are the ones people liked. A shape shared by the survivors and by the casualties alike cannot by itself explain why the survivors are admired. Braithwaite does not claim that the geometry is irrelevant; she claims that a list of famous halls is the output of a century of filtering and cannot be read as though it were the result of an experiment.

EIn 2021 Vermeer and Duarte tested the disagreement together. The Kestrel Bay Concert Hall, a vineyard-plan room opened in 2004 and criticised ever since, was fitted with movable reflecting panels along its flanks; deploying them raised the lateral energy fraction at the surveyed seats from 0.13 to 0.22. Four hundred subscribers rated eight concerts across a season, the panels being deployed or withdrawn at random and the audience not told which arrangement was in use. On the question of envelopment — whether the sound seemed to arrive from all around — the treated concerts scored significantly higher. On overall enjoyment the two conditions were indistinguishable. Vermeer takes this as confirmation that the quantity behaves as her model predicts. Duarte takes the same figures to show that a property an audience can detect but does not care about cannot be the property that makes a hall admired. Braithwaite regards the experiment as informative about one modern room and silent about the historical question she is asking.

FUnderneath the argument about walls lies a harder problem about proof. A hall cannot be randomised: nobody builds forty rooms and assigns orchestras to them at random, and no listener can hear two halls at once, so every comparison is made against a memory weeks old. The difficulty presses unequally. It presses least on Vermeer, whose central quantity is a ratio of energies that a microphone records without asking anybody a question, and hardest on Duarte, whose entire case is assembled out of preferences that people report. There is also a dispute about the word 'good' which the three rarely make explicit. Vermeer uses it of a room, as a property present whether or not anyone is listening; Duarte uses it of an occasion, as something an audience produces on a particular evening; Braithwaite uses it of a reputation, the verdict a city reaches over generations and then stops re-examining. Three quantities wear one adjective, and the evidence for each keeps arriving as the answer to somebody else's question.