the strained shape. Mukhophadhyay (1973) argued that quartz
grains in the Ardennes are good markers because they are sepa-
rated enough not to interact with each other. This is intuitively
reasonable. Markers used here, however, all show some
interaction.

Impingement of grains upon each other produces elonga-
tion of the grains or one of the grains. Elongation is always
in the plane of stylolitization which corresponds with the direc-
tion of secondary mica growth (ie. the cleavage plane). Grain
impingement causes deviation from the strain path of an ellipse
by imposing additional grain shape changes. These grain inter-
action effects would be small if there were no grain boundary
sliding, or viscosity contrasts, because bounding areas would
deform with the clasts but would always retain their configur-
ation. Two touching clasts would remain just touching through-
out the deformation. Within the studied suite of rocks viscosi-
ties differ between the clasts, and there is probably grain
boundary sliding. Grain interactions do occur and, along with
the initially irregular grain shapes, cause final strain marker
distributions to have features which are not predicted by
geometric deformation.

Viscosity contrasts

Clasts of all compositions were measured and included together
in most of the strain determinations.

For discussion, let us assume a perfectly random distribu-
tion of heterogeneous clasts. Clasts that are more viscous than
the matrix will display a component of rigid body rotation when
strained (Gay, 1968a, p. 219). Gay (1968a) pointed out that
clasts, with viscosities much greater than the matrix, will exhibit
little shape change, but will undergo rotation such that their
long axis will coincide with that of the strain ellipse. Clasts,
which are nearly equivalent to the matrix in viscosity, will
deform geometrically and not rotate rigidly, thereby approxi-
mating more closely the deformation predicted by geometric
strain analyses.

Clasts with moderate to low viscosity ratio will approximate
more closely the deformation predicted by geometric analysis
of the strain. Their ellipticities will generally be greater than
for the high viscosity ratio clasts. On the polar plot the distri-
bution in a strained state would be directionally uniform, but
display a spread in magnitude of final ellipse ratios. This would
have the effect of inducing bimodality in the final ellipse dis-
tribution on the polar plot, if the strain ellipse were taken to
be the central part of the distribution. A characteristic of this
type of induced biomodality is that the strain ellipse axes will
parallel the two directional modes of the unstrained distribu-
tion. Gay (1968a) assumed that the boundaries between matrix
and clast are fixed (ie. there is no creep or slippage). In the pres-
ence of pressure solution phenomenon this assumption would
not hold.

A consequence of this assumption being invalid is that the
rigid body rotation component of competent clasts need not
occur. Gay (1968a) pointed out that clast interference has a ten-
dency to lessen the control of viscosity contrast and therefore
presumably decrease the effects of rotation.

Figure 61 illustrates bimodality of a distribution produced
by viscosity contrast between clast and matrix. The distribu-
tion tends to be bimodal where only quartz clasts are involved
and even more accentuated where less viscous clasts are also
present.

Figure 61. These are polar plots of the natural log of
the axial ratio and orientation of the long axis of the final
ellipses from sample 1. Plot A contains data from quartz
grains and plot B combines these with data from chert
grains. The difference in dispersion of the pattern from
the origin is because the chert shows more deformation
than the quartz. These differences are due to viscosity
contrasts between the grains, and between the grains and
the matrix.

95