TUNES 45

possible to put any ballad tune on a piece of paper not larger than
2 in. by 2 in., that is well within the smallest cards used in card-
indexing. Thirty or forty such transcriptions can be before our
eyes if we wish to compare tunes, replacing advantageously the
one or two ponderous tomes of staff notation which are otherwise
the most feasible. Also no one who has learned an alphabet can
fail to grasp the notion of notes different in pitch, even if he were
tone-deaf. To characterize a tune, we have to remember that
ballad tunes vary, without losing identity, in key, pitch, rhythm,
and ornamentation. All that is constant is the melodic curve, or
significant parts of that curve. This curve makes a graph on the
lines of the staff notation, but the graph may be represented quasi-
algebraically, by using numbers for rising semitones, and letters
for falling semitones. We need more than ten rising semitones,
since there are twelve in an octave, so our numbers must be drawn
from different fonts of type: roman and italic, or black-ribbon and
red-ribbon on the typewriter. For most purposes of comparison
the melody becomes clear in the first phrase of a ballad tune, so
that it is enough to record it so. Identity of formula is very rare
without identity of tune. The first phrases of the French Roi
Renaud and En passant par la Lorraine happen to coincide in yzD
(i.e. a rise of seven semitones, rise of two, fall of four); the com-
plete transcriptions indicate differences which seem to exclude the
chance of an association between the tunes.1 On the other hand,
differences in the melodic contour do not necessarily sever the con-
nexion between two tunes, since the difference may be super-
induced on the original tune by some sort of ornament. Such
ornaments are wont to reveal themselves in that they leave sections
of the original tune in the same relative positions, that is, the orna-
ment returns the same distance that it diverges. So, for two tunes
of Clerk Colvill with formulas EC^zzEE and BC32BC52BB, we
find the whole of the first in the second, with the addition of an
inserted BC5 (i.e. five semitones down and five up). Of course, as
we are measuring intervals only between one note and another, it

1 See Th. Gerold, Chansons populates des XV& et XVI6 siecles (Bibl. Roman.),
Strasbourg, n.d., Appendix 3, 7. Taking the first two lines of the former, which
is in quatrains, and omitting the refrains of the latter, one may accentuate the
similarity of the two tunes, without erasing the differences, thus:
Roi Renaud. En passant.

Jjto68U8 j flat 68 US

ddd/a.. a6£/g.. ccc/d.. cd5/a. r f. f/c. c\c.c/d.bj c. elf. c/c. b/a. g