Shaw first reported on Brahms's Requiem in an article called ‘Gas and Gaiters’ which appeared in The Star in May 1890. ‘A solid piece of music manufacture … it could only have come from the establishment of a first-class undertaker.’ The tedium of sitting through it had been intolerable, and for years the memory haunted him: ‘There are some sacrifices which should not be demanded twice of any man; and one of them is listening to Brahms's Requiem.’ The experience, he said, ‘is patiently borne only by the corpse’.

This, I had always thought, was Shaw being Shavian. In fact there were plenty who shared his boredom if not his happy way of expressing it. The Musical Times, writing on the first London performance in 1873, detected ‘a feeling of weariness in the audience’, and the Monthly Musical Record, which was of opinion that ‘Herr Brahms is a very unequal composer’, felt that the work would ‘have certainly gained in effect by judicial curtailment’. What has surprised me is to find the sentiment echoed at this very day by friends and colleagues who, as they say, ‘cannot stand it’. The precise grounds for criticism or disaffection have not been made clear but, having listened to the work in nearly thirty recordings, I can certainly see one possible cause.

Elijah has been somewhat out of favour now for a number of years. The seemingly mild treatment of the biblical story has counted against it in an age which prefers something bolder than Mendelssohn's benign idiom. But, though the work may be overlong and some of its choruses excessively anodyne, it has a certain conviction and character all its own. Our problem may be that we are hearing a work written in 1844 through ears that have encountered all the Victoriana Mendelssohn engendered, but the kind of music he was composing was, at the time of its writing, comparatively new in idiom: Bach and Handel heard, as it were, through the light of mid-nineteenth-century harmony. However that may be, Elijah himself, in sure hands, emerges as a real, strong character, at once vehement, despairing and humane, and the rest of the solo writing is on the highest level of Mendelssohn's achievement. Besides, when the work is sung in the language (German) for which the notes were composed and with really purposeful and clean choral singing, as happens on the Sawallisch set, recorded in Leipzig with its genuine Mendelssohn tradition, his most favoured oratorio emerges as a picture newly cleaned, alive and immediate, not least because Sawallisch follows Mendelssohn in assigning the passages for concerted solo voices to these and not, as happens in most other recordings, to a boys' choir or to a reduced chorus.

The libretto is largely the work of the composer's theologian friend, Paul Schubring, who had helped him with St Paul.

Ideas about the performance of early music have changed radically over the past forty years. There has been considerable research on the vocal and instrumental forces Monteverdi is likely to have had in mind. For most baroque music, we now prefer an athletic and transparent texture to the lush, full sound of yesteryear. Different views of tempo go hand in hand with these ideas; in fact, in some ways they flow almost inevitably from them. Those in their turn lead to a questioning of earlier assumptions about form and design. Pitch, pitch-standards and intonation have similarly been examined exhaustively in a historical context, with results that have a fundamental impact on the performer's approach. In the particular case of Monteverdi's 1610 Vespers there has been enormous discussion of how far it is really to be seen as a single work, whether liturgical material needs adding, and if so how much. Few of these areas of discussion have led to unanimously accepted conclusions: that is often the way with historical investigation, in music as much as in anything else. And not all conductors have shown themselves equally informed or equally prepared to accept the latest conclusions at any one time. But the range of areas in which attitudes have changed means that each performance is to some extent a child of its time. The date of a recording is important.

Not that the date of itself says much about the quantity of musical pleasure to be derived from a particular recording. Most musicians would accept that there are qualities of musicianship which retain their power irrespective of historical purity.

Janáček was entering his seventy-third year when he composed the Mša Glagolskaja in 1926. Long before, he had composed one Mass, now lost, and begun another which he left incomplete. The ‘Glagolitic’ is one of the fruits of his famous Indian summer, warmed by his passionate affection for Kamila Stösslová, thirty-eight years his junior. He had just finished liis Sinfonietta, an instant success, and more music was still to come from his remaining two years: the Second String Quartet, the Říkadla (‘Nursery rhymes’) settings, the Capriccio for one-armed pianist and ensemble, and the opera From the House of the Dead.

By Janáček's advanced standards of quirkiness, the Mass is rather straightforward – wholly idiosyncratic, of course, but nothing like so odd as the Capriccio or the last opera or quartet. There is an obscure quirk in its title: having chosen to set the Mass not in Latin but in an Old Slavonic version from the ninth century, he seems to have got the idea that the word ‘Glagolitic’ applied to that language. In fact it means the script in which it was originally written down, with which of course he had no concern. Certain sections of the Latin text are shortened in this Slavonic Mass; the Sanctus and the Benedictus are run together. To the choral movements, which include all the solo singing as well, Janáček added not only an Introduction but a final Intrada for orchestra, and between the Agnus Dei and the latter a churning Postlude for organ alone.

In the great concluding achievement of his career Haydn looks back and celebrates his love of nature and his deep sense of man's harmony with the natural world. As Robbins Landon remarks, The Seasons is a valedictory work; Haydn is ‘describing things – the first glow of dawn on the horizon, the hazy heat of midsummer, the ripe fields of autumn – which will soon retreat beyond his grasp’. Yet they are evoked with all his former energy and zest. Though Haydn complained that the task of setting Baron van Swieten's text exhausted him, there is little sign of it in the music. Age left its mark not on the freshness and vigour of his creativity but only on the character of the work, which culminates in the wonderful bass meditation on the brevity of human life, followed by the jubilant final chorus expressing the composer's sure faith in the eternal life to come.

It may be that Haydn would have liked van Swieten to have given him more opportunities for this sort of thing, as he had done in The Creation (on which the two men had collaborated a few years earlier), and less of what Haydn himself disparaged as ‘Frenchified trash’ – the musical imitation of frogs and other fauna and flora which he said was forced on him. Haydn also said that The Seasons was inferior to The Creation because it was about peasants whereas The Creation was about angels. But The Creation too has its roaring lion, its ‘flexible’ tiger and its crawling worm.

For many Gerontius is Elgar's outright masterpiece. Having completed the work in the summer of 1900, he commented: ‘The trees are singing my music. Or have I sung theirs?’ So he himself sensed that the work had a special power and beauty, confirmed when he signed off the score ‘This is the best of me’. As David Cairns has remarked: ‘No other work of his carries so strong a sense, in a good performance, of having been given, of coming from deep and ancient wells of experience.’ There is something inevitable in the setting of Cardinal Newman's text that suggests the work has somehow been there since time immemorial, and that setting itself, combining parlando and lyrical outpouring as appropriate, provides a style perfectly attuned to the deeply felt emotions of the words. A devout Catholic, Elgar responded to the mystical sincerity of Newman's vision with his own. At the same time the work is as close as Elgar came to writing opera. In a convincing performance, the dramatic quality of the writing should carry all before it. And all the recorded accounts of the work are convincing because artists seem to respond instinctively to the greatness in the writing and give of their very best.

That is true even of the very early (1924) version on Edison Bell, although this is little more than a curiosity. The work is substantially foreshortened, the choruses are sung by a choir of just eight solo singers, and the orchestra consists of just twenty-four players! The results are, to say the least, not very satisfactory. But these seven 78rpm discs are not without interest.

(Numbers in brackets refer to corresponding numbers in the discography)

The history of the St John Passion is more complex than that of Bach's other surviving setting of the Passion story, the St Matthew Passion (qv). First performed in 1724, three years before the first version of the Matthew, this direct and deceptively ‘simple’ oratorio was subjected to a revision the following year that resulted in the substitution of new choruses for the opening and concluding ones, and the insertion of three alternate arias in the body of the work. All of the music for this first revision survives.

In the early 1730s, Bach returned, in essence, to the 1724 sequence, but with a few modifications. In the intervening years, the ‘new’ opening chorus, ‘O Mensch, bewein’ dein Sünde gross', became the concluding chorus of Part One of the St Matthew Passion, and thus was no longer available for use in the St John. Furthermore, ecclesiastical authorities in Leipzig had evidently objected to Bach's insertion of two intensely dramatic sequences from the Gospel according to St Matthew into the St John Passion (hereinafter SJP), and he removed them. Bach provided no replacement for the first of these excised interpolations, the passage describing Peter's remorse at his denial of Christ; but for the second, the earthquake episode after the Crucifixion, he substituted an instrumental sinfonia that has not come down to us. The aria that he wrote to replace ‘Ach mein Sinn’ in this third form of the SJP has also not been preserved. Finally, this third version did not have the chorale that follows the concluding chorus in the first version.

The Missa Solemnis is exceptional among great choral works in its dependency upon a great performance. Byrd's four-part Mass can be sung by anxious amateurs and still move its hearers. Mozart's Requiem is given every year by countless school choirs and choral societies and still manages to impress. Bach's Mass in B minor would sound like great music even if sung by a few sopranos and a bass gathered round the piano on an old-fashioned ‘musical evening’ at home. But the Missa Solemnis is different. It makes almost superhuman demands of its performers, and the best that they can give will not amount to much unless an outstanding conductor is there to control the massive forces, give them a vision and inspire them. Of the orchestra it demands unrelaxing attention to details of rhythm, accentuation and balance. The five soloists (four singers, one violinist) have immense opportunities, including (for the violinist) the equivalent of a concerto's slow movement, but the responsibilities are correspondingly awesome. For the soprano soloist in particular, Beethoven's demands upon high notes and breath-control are almost extortionate. The greatest challenge of all confronts the chorus. No choir can attempt the Missa Solemnis without a strong tenor line: that sets a formidable condition for a start. By the end of the evening all voices are likely to feel sung out. ‘One more page and I'd have died’ they say to each other when it is all over. Again the soprano part is cruellest. Beethoven thinks nothing of sending them up to the B flat, there to stay singing ‘Et vitam venturi’ till they begin to wonder whether the life to come may not be more imminent than previously supposed.

After the injection of post-Wagnerian Romanticism administered to English choral music by Elgar through The Dream of Gerontius in 1900, four major works appeared during the subsequent sixty-two years which showed how powerful a stimulant it had been to composers' imaginations. The first, Vaughan Williams's A Sea Symphony, was heard at the Leeds Festival in 1910 and most obviously betrays the Elgar influence while at the same time proclaiming a new and powerful individual voice. Of the other three, Walton's Belshazzar's Feast was composed between the wars in 1931; Tippett's A Child of Our Time appeared during the Second World War, in 1944; and Britten's War Requiem was first performed in 1962, seventeen years after the end of the Second World War, although it is the First World War that it commemorates. Of these four works, only one, Britten's, may be said to be religious, although its use of the Latin text of the Requiem Mass is almost secularised by the interpolation of nine poems by Wilfred Owen. It is true that Belshazzar's Feast has a text compiled from the Old Testament by Osbert Sitwell, but it is not religious music in any spiritual sense, while Tippett's oratorio A Child of Our Time has a libretto by the composer which is interspersed, in the manner of the chorales in Bach's Passions, by Negro spirituals. Vaughan Williams selected passages from Whitman for his Sea Symphony, some of which have a religious connotation.

Belshazzar's Feast is the most frequently recorded of these works, and probably the most frequently performed too.

Mozart's last, sombre masterpiece, the Requiem in D minor K.626, has caught the public imagination as few other works, largely because of the mystery surrounding its origins.

The bare facts are now known: Count Walsegg von Stuppach, a passionate music-lover, commissioned the Requiem in the spring or summer of 1791, in memory of his young wife who had died earlier that year. Walsegg was a harmless eccentric whose practice it was to recopy works he had commissioned, in order to test his chapel-musicians' powers of deduction, and who seemed to derive much innocent amusement when flatteringly credited with their authorship. Hence his insistence on retaining the copyright of the Requiem so that it could not be published prior to his own first performance, thus spoiling his little joke. The contract between Mozart and the Count has recently been discovered, proving that their business together was above board and conducted with total propriety.

Nevertheless, the legends persist, and even today commentators are divided over the more colourful theory that Mozart had a premonition of his impending death, regarded Walsegg's ‘grey-clad emissary’ as a messenger from the other world, and became obsessed with the idea that he was composing the Requiem for his own funeral. Though a letter in which Mozart confides these fears to Da Ponte has long since been discredited as a forgery, it is significant that the account given by his widow Constanze to contemporary chroniclers seems to corroborate this view.

In the event, Mozart did not live to finish the Requiem. This placed Constanze in a delicate position, since a large amount of money had already changed hands, and more was outstanding subject to the Requiem's completion.

Both of Stravinsky's largest choral works, Oedipus Rex (1926–7) and Perséphone (1933–4), fall outside the scope of this book, the one because, as an ‘opera-oratorio’, it has already been covered in Opera on Record 2 (Hutchinson, 1983), the other more mundanely because there have been too few recordings to form a basis for comparison. The same reason of neglect disqualifies other pieces, including the great sequence of late choral monuments: the Canticumsacrum (1955), Threni(1957–8), A Sermon, a Narrative and a Prayer (1960–1) and the Requiem Canticles (1965–6), of which the last was one of the few works for large forces that the composer himself did not record. But there is a typicalness too in Oedipus Rex in its straddling of worlds, its being at once a drama, a concert item and a ritual. The Wedding (1914–23), for example, was composed as a ballet of a ceremonial nature, but can work in the concert hall as a ritual cantata. And the more straightforwardly sacred pieces, beginning with the Symphony of Psalms (1930), are liturgies transposed on to the concert platform: apart from small unaccompanied prayers, only the Mass (1944–7) sets a standard liturgical text, and here the accompaniment for ten wind instruments makes concert performance more likely. All Stravinsky's choral works are at once sacred and secular, theatrical and purely musical. So it is entirely suitable that his two most frequently recorded compositions of this kind should be a ballet of wedding songs and a symphony of psalms.

The Wedding (Svadebka; Les noces) had a more protracted genesis than was at all usual for Stravinsky.

The published text of Fauré's Requiem is a strange and contradictory document. On paper the work looks as though it is scored for a standard orchestra, but closer inspection reveals how very curiously used the instruments are. There is not a single bar for full orchestra. The violins are undivided into the customary firsts and seconds, they play in only four of the seven movements and often do no more than double one of the viola lines (Fauré suggested that the viola parts could be enriched by urging some of the better violinists to switch to violas during the violin-less movements). Flutes and clarinets (always doubling strings or organ) are heard only in the Pie Jesu, the second clarinet for less than four bars; timpani and trombones (doubling the horns in all but eight bars) appear only in the Libera me. The woodwind, in particular, are not so much restrainedly as invisibly used. The question of how this anomalous score should be performed can hardly be discussed without considering how it came into existence.

Fauré began writing the Requiem in 1887 and its first version was ready in time for the funeral of a certain M. le Soufaché at the church of the Madeleine in Paris, where Fauré was choirmaster, on 16 January 1888. He worked unusually fast, possibly under the impact of the illness and death of his mother: the manuscript of the Agnus Dei is dated 6 January 1888, less than a week after Hélène Fauré died, that of the Sanctus three days later.

For a professed agnostic, if not an atheist, Berlioz shows a remarkable interest in sacred music. A significant proportion of his output is concerned with religion – and this in a period when sacred music was in decline in France. Three of his major works – those dealt with in this chapter – are religious works; and in each case the original impulse came from within, from the composer himself, not from a commission. Even in his secular works one is struck by the frequent recurrence of religious imagery. This is more than just a reflection of the Romantic vogue for religion as a picturesque detail of landscape: it is a preoccupation.

Yet the paradox is easily explained. ‘For seven years’, wrote Berlioz of his boyhood, religion was ‘the joy of my life’. The very loss of that joy left a deep imprint on his consciousness, and his music reflected it. It is not the music of a believer; but it conveys an intense regret at the inability to be one, and a profound awareness of the need, the desperate need, to believe, to worship. The Requiem, or Grande Messe des Morts, evokes the possibility of a universe without God; the God of the Te Deum is more terrible than compassionate; L'Enfance du Christ looks back nostalgically to a time when the healing myths of Christianity were received unquestioningly and nourished all.

Berlioz's own lack of belief is, as it were, used to suggest the eternal hopes and fears of the human race, faced with the enigma of death; his intuitive understanding of the religious instinct, his unsatisfied yearning for faith, enable him to respond to those immemorial feelings and express them in his art.

Even more than Verdi, Rossini has suffered from the prejudice which would have us believe that the sacred music of an Italian opera composer will be compromised both by its worldliness and by its lack of real scholarship. Down the years, his not entirely well-founded reputation as an unschooled, cynical, bon viveur has made it difficult for the listening public and a good many performers to treat his sacred music with the consideration and care it invites. On record, the result has been depressing cycles of performances that enact and re-circulate assumptions about Rossini's competence and sincerity that bear little relation to the letter or the spirit of the two fine sacred works he wrote in the years following his retirement from theatre composition.

When the Abbé Gallet de St-Roch put the usual ritual questions about belief to the dying Rossini in November 1868, Rossini is said to have replied: ‘Would I have been able to write the Stabat mater and the Messe if I had not had faith?’ Outwardly, he was not a religious man; his faith was neither formal nor institutionalised. Privately, he might well have nodded agreement at the words of his famous contemporary, the poet Tennyson, in his In Memoriam:

The spectre of doubt, expressed as a kind of serene anxiety, is certainly there in parts of the Petite messe solennelle, written in 1863–4, not least in the work's quizzical conclusion. Equally, Rossini was a man in whom the primary affections and loyalties, both sacred and familial, were strong; and it is these primary forms of belief that have often inspired some of the most personally affecting religious art.

One must begin with f, and then tune its octave pure. After that tune c′, a fifth from f, and make it completely pure. Then lower it just enough that it still seems good and the ear can tolerate it. From c′, tune its lower octave pure. Then tune its fifth g in the same way, narrowing it to the same degree as the first. Then tune its upper octave pure, which is g′. Tune d′, and then tune this fifth in the same way, narrowing it like the others. Then pause at this point and perform the trial, which is done in this way. Tune b♭, next to c′, to the fifth f′, next to g′, and keep the b♭ a bit high, so that this fifth is tempered [temperée] and is the same as the others. Then play the d′ that you have tuned, which makes the major third against b♭ and the minor third against f′. When this chord is found to be good, everything which has been tuned is good, because the tuning is proven only by the thirds. When they are found to be good throughout, the tuning is correct.

Continue then, and follow the order used at the beginning. Go by octaves and fifths to the last note, and do not tune any fifth after the first trial mentioned above if the third within it does not prove to be good, as you may observe in the example that follows.