First, let’s look at the Part I data. You can see that a greater percentage of long-held thirds trended smaller when compared to the passing thirds. In part one, the players sustained the long-held notes for two to three beats at an andante (medium) tempo. However, the passing thirds had only one beat.
For the Part II data set, the players sustained the chords for a much longer time. They held these harmonies for four to six beats. You can see that the major thirds and tenths in Part II trended smaller much more often than in Part I.
Keep in mind that the violin players used vibrato on all of the notes. This means that the long-held thirds trended toward ideal despite the heavy vibrato. With vibrato absent, the trend toward ideal thirds would likely have been much greater.
Shackford concludes:
The major third in measure 5 is on the small side, at one point being actually the size of the 4/5 interval of the harmonic series. This is consistent with the data… in Example 20 (part I) which showed major tenths becoming smaller as they were held, one actually nearly reaching 4/5, and indicates that in slow-moving, sustained music the acoustical nature of the tonal material of music is an important factor in determining pitch and harmonic interval size. (Shackford 2:80)
Stating that a major third has a 4/5 ratio is the same as saying it has a size of 386 cents. Basically, he says here that a harmony’s length impacts playing (or singing) in tune. In fact, how long you hold a chord has a massive effect on its tuning.