I must pitch into Marriner, as I have many conductors before him, for not observing the exposition repeat in the first movement of the Italian. this is not a bee in my bonnet: it is a matter of plain musical commonsense. I do not quarrel with the similar omission in the Scottish, for Mendelssohn only writes three perfunctory bars of a repeated chord to lead back to the key of the opening; but in the Italian he composes about 22 bars of lead-back, new and delightful music which he clearly intended to be heard, about the most charming da capo you could imagine. How often has anybody heard them? Well, you can hear them on the alternative DG recording by Sinopoli.
Marriner gives an excellent performance of the Scottish. His opening tempo is pretty slow for an Andante con moto (crotchet = 60 rather than the composer's 72) but it does convey the memory in Mendelssohn's mind of the ruined Abbey of Holyrood, and gives an atmosphere that is continued at the start of the Allegro till the piu mosso breaks out and, later, the 'storm'. The rest is fine, with particularly splendid playing in the scherzo and, most important, a coda to the finale which is as buoyant as I have ever heard.
The Italian again shows the skilful playing of the ASMF's strings pp and, while the tempo of the opening is pretty swift, it does not lack a feeling of expansiveness (which Sinopoli's does). The middle movements are well contrasted (both are very slow from Sinopoli) and the Neapolitan procession is a fairly lightly-stepping one. Sinopoli is very lugubrious (though his basses with their far longer staccatos do sound appropriate to a church procession—Marriner's might almost be mistaken for pizzicato). Sinopoli's articulation and phrasing in the third movement is most careful, while his finale is also good, though it does not quite keep up the pressure when the 6/8 changes into common time.
The recorded balance is strikingly different between the two, with the lower instruments more heavily recorded from Sinopoli than from Marriner. The timpani, curiously for CD, are not notably clear and defined on either recording.'