The problem is—for reasons Verdi touched on when he enunciated the work's several distinguishing features—Il barbiere is an extraordinarily difficult work to bring off on record. In the theatre it gets by because Beaumarchais, Sterbini and Rossini have written a show that is largely irresistible, horribly so in places. However inept the staging or the singing—in the past it was often the latter, nowadays it is usually the former—the piece survives, and the punters go home happy. On record, though, it stands or falls by the singing (the writing by turns declamatory and virtuoso), the conducting, and the degree to which everything comes together in ensemble playing that has a feel for the 'run' of the music.

It is a sign of how much progress has been made in the last 30 years in the technically difficult business of singing Rossini (Urtext Rossini, without cuts) that EMI and Teldec can assemble such strong rival casts. (See EMI recording) True, by some contractual sleight-of-hand Samuel Ramey, a superb Basilio, appears to be able to play for Juventus one day, Real Madrid the next; but that is the only surprising overlap. Choosing between Hampson's Figaro (EMI) and Hagegard's (Teldec) is well-nigh impossible. Hampson is the suaver of the two, Haggard rather more obviously hail-fellow-and-well-met. The same could be said of the two Almavivas. It is arguable that Gimenez is the better Rossinian, the more practised virtuoso; but it is Hadley who sings his minor key aubade as Rossini directs, meltingly a mezza voce. Corbelli's Dr Bartolo (Teldec) is masterly, but Pratico (EMI) is just as good in his way. Nowadays, Dr Bartolo seems to be thriving; Lucio Gallo's performance for Abbado is one of the obvious highlights of the DG set.

It is also virtually impossible to get anything much thicker than a piece of rice paper between Susanne Mentzer's Rosina (EMI) and Jennifer Larmore's (Teldec). Both are technically superb. What, though, of characterization? Mentzer and Larmore tend to take the colour of their readings from the conducting, from the ambient atmosphere. So, to some extent, does Berganza on the earlier of the two Abbado recordings. Callas (EMI/Galliera) and de Los Angeles (EMI/Gui) are utterly different, not only in temperament but in style. I would not willingly be without either though Callas's Rosina was never popular with Italians. ''She wasn't Rosina, she was Carmen!'' Thus, Carlo Maria Giulini—surprisingly sharp and decided—in a conversation I had with him in Milan recently. (For him Berganza is the ideal.)

Which brings us to the two conductors Gelmetti and Lopez-Cobos. They could not be more different. Lopez-Cobos's conducting is quick and cool. He gives us Rossini the dispassionate ironist. Gelmetti, by contrast, is far more searching. Gelmetti treats Rossini like royalty, like Mozart (a slightly romanticized Mozart, it should be said). After Gelmetti and his stylish band of Tuscan instrumentalists, Lopez-Cobos sounds rather cold-blooded. And this sometimes spreads to his singers as they fire off their vocal salvoes with deadpan accuracy. Yet by delving beneath the surface Gelmetti creates problems of his own; in particular, he often loses sight of rhythmic continuity and the work's larger comic rhythm.

Gelmetti writes at length on this and other subjects in EMI's booklet, though, tactfully, EMI have confined the more convoluted sections of his essay to the Italian note. (The main essay, which I wrote, has one of Gelmetti's more persuasive paragraphs as its epigraph.) Teldec, by contrast, have assembled an anthology of contemporary comments—Stendhal and the like—including a letter in which Rossini talks of the primacy of rhythm, as well as making an astonishing attack on Beethoven. Unfortunately, the letter is bogus. Nowadays recordings use good musical texts. Equally, booklet editors should use comparably up-to-date sources—in this case the Fondazione Rossini's latest publication Gioachino Rossini Lettere e Documenti Vol. 129 Febbraio 1792–17 Marzo 1822 (Pesaro: 1992)—before foisting these tatterdemalion titbits on a gullible public.

Faced with a pair of—in many ways fine—complementary sets, the reviewer's usual refuge is ''you must have both''—a reprehensibly idle piece of advice to those of us who have wives and children to feed. From the economic point of view the two-CD Teldec set is clearly the better buy, though there is a disc change to be negotiated in the Act 1 finale and there are only half as many cue-points as on the elaborately edited and absolutely complete three-CD EMI set.

The other reviewer's refuge—stay with a tried and trusted favourite (in my case, the Gui set)—is less reprehensible, though it sidesteps the kind of advances there have undoubtedly been textually and technically in the last 30 years. My problem with the two newcomers is that, at the moment, the brilliantly consistent Teldec tends to leave me cold, whilst the rather more wayward EMI already invites replaying at various moments. How gorgeously, for instance, Gelmetti's Tuscan strings colour the legato crotchets in the Act 2 shaving scene as Almaviva and Rosina plan their midnight tryst. This is real love music. Whoever said Rossini lacked heart?

-- Richard Osborne, Gramophone [11/1993] review of complete opera Teldec 74885