Superposed Deformations and Their Hybrid Effects

Journal of the Geological Society, London, Vol. 160, 2003, pp. 117–136. Printed in Great Britain.

117

Superposed deformations and their hybrid effects: the Rhoscolyn Anticline

unravelled

SUSAN H. TREAGUS, JACK E. TREAGUS & GILES T. R. DROOP

Department of Earth Sciences, University of Manchester, Manchester M13 9PL, UK (e-mail: [email protected])

Abstract: This study of the controversial structures of the Rhoscolyn Anticline suggests a different result of

two-phase coaxial deformation from Ramsay’s Type 3 interference fold patterns. From detailed field

observations of the sequence of bedded quartzites, psammites, pelites and oblique quartz veins, with their

strong competence contrasts, we conclude that the Rhoscolyn Anticline was an original tight, upright F1

anticline that has undergone modification and distortion in a second deformation (D2). This second

deformation is an oblique, but near-vertical, pure shear, with a quantifiable strain ratio (R ¼ 3) that altered the

Rhoscolyn Anticline and its minor structures into a more open, SE-overturned antiform, with c. 260 m hinge

migration. Refolded folds are rare, but hybrid F1 þ F2 minor folds and their fabrics, especially in the region

between old and new hinges, provide clues to the two-stage history. Oblique distortion of originally NW-

verging F1 minor folds has resulted in their apparent neutral vergence in the present-day hinge of the

Rhoscolyn Anticline. We regard the structures and fabrics in quartzites and psammites as more reliable

indicators of the region’s deformation history than those in pelites or quartz veins, and this may prove true for

other regions of polyphase deformation.

Keywords: Anglesey, polyphase processes, superposed deformation, folds.

Observations of geological structures such as folded fabrics,

crenulation cleavages and folded folds, lead to the conclusion

that a region has undergone polyphase deformation. The nature

of this polyphase deformation can be understood from the 3D

geometry of refolded folds and their sectional interference

patterns (Ramsay 1962; 1967, p. 531; Ramsay & Huber 1987, p.

492), and from the geometry of multiple deformation fabrics in

the field or thin section, such as crenulation cleavages (Passchier

& Trouw 1996, pp. 84–88) or folded lineations (Ramsay &

Huber 1987, pp. 481–484). In this paper, we consider an area of

polyphase deformation whose structures have led to the emer-

gence of many different interpretations, with significant implica-

tions for the tectonic history of the region.

Our study focuses on the Rhoscolyn Anticline located on Holy

Island, Anglesey (Ynys Mon), a 2 km2 coastal area of Monian

Supergroup rocks that is a popular place for teaching elementary

mapping of distinct lithologies around a major fold structure, but is

equally useful for investigating the structures of polyphase defor-

mation from the large to small scale. The area was chosen by Price

& Cosgrove (1990, pp. 482–490) as a case study for structural

analysis of multiple deformation, and by Lisle (1988) to question

the application of vergence principles in refolded regions. The

structures in this area have given rise to many different interpreta-

tions, some of which are illustrated in Fig. 1 (Greenly 1919;

Shackleton 1969; Cosgrove 1980; Phillips 1991); other interpreta-

tions include those of Barber & Max (1979), Lisle (1988), Roper

(1992) and Hudson & Stowell (1997). Many of these workers differ

in their interpretation of the number and significance of the

deformation phases that gave rise to the fabrics and folds. In

essence, these simplify into whether the Rhoscolyn Anticline is a

major first fold, as first proposed by Shackleton (1969) (Fig. 1d),

that is overprinted and modified by later folding and fabrics, or is a

later antiform, that refolds earlier structures, as exemplified by the

other illustrated interpretations (Fig. 1c, e and f).

The Rhoscolyn area provides ample exposure of the follow-

ing types of structural criteria, which have been used variously

to back the different interpretations. Major and minor folds

with clear vergence relationships change their sense around the

Rhoscolyn Anticline. Foliations vary from grain-shape fabrics

in quartzites and psammites, to a dominant crenulation clea-

vage in pelites, with intermediate rocks showing folded

cleavage within a bed, and sometimes two cross-cutting

cleavages. Folded quartz veins are abundant in pelite beds, and

roughly track the first cleavage. Despite all these features,

textbook-style coaxial refolding patterns of Type 3 (Ramsay

1967) are rare, leading us to wonder whether coaxial refolding

is the primary method of superposed deformation in these

rocks. Critical questions in the field, here, are whether folds on

different scales are ‘first’ or ‘second’, whether their vergence

is significant, and how the different cleavages in different rock

types relate to their folds.

We will propose a two-phase model that can account for many

of the ambiguities in this area, and reconcile some of the

differences among previous interpretations outlined more fully

below. Our model and structural observations are restricted to the

tripartite South Stack Group, and we leave investigation of the

differently deformed overlying New Harbour Formation for

inclusion in our continuing investigations of the geology and

structure of NW Anglesey. Our approach begins by reviewing

mechanisms of superposed deformation and polyphase folding,

with special focus on coaxial refolding in rocks with competence

contrasts. This leads to our specific investigation of the structures

associated with the Rhoscolyn Anticline, and the development of

a quantitative model for the two-phase deformation and folding

history. Our model has implications in general for polyphase

deformation in metasedimentary rocks with competence con-

trasts, and for the regional tectonic history of the Monian

Supergroup of Anglesey.

Fig. 1. (a) Regional setting and (b) summary map of the Rhoscolyn Anticline, Holy Island, Anglesey, and the principal previous interpretations and

polyphase history in schematic cross-sectional form of (c) Greenly (1919), (d) Shackleton (1969), (e) Cosgrove (1980) and (f) Phillips (1991). In (c), (e)

and (f), the left-hand diagrams illustrate ‘D1’ stages. B, Borthwen; BD, Bwa Du; CG, Coastguard; PS, Porth Saint; v, volcanic rocks; Q, quartzite.

S .H. TREAGUS ET AL .118

Mechanisms of superposed deformation and coaxialrefolding

Refolded folds are probably the least-disputed evidence for

polyphase deformation. Fold interference patterns in map or

section view (Ramsay 1962, 1967, chapter 10; Ramsay & Huber

1987, chapter 22; see also Thiessen & Means 1980) reveal the

variety of geometry that can arise from two phases of folds,

according to the mutual relationships of their axes and axial

planes. Ramsay’s classification (Ramsay 1962, 1967, chapter 10;

Ramsay & Huber 1987, chapter 22) of fold interference (Types

1–3), is based on two superposed phases of similar folding with

the same wavelength and amplitude, and has been illustrated by

spectacular examples from gneissic rocks.

In this paper, we are interested in the processes of two coaxial

phases of folding, as for the Rhoscolyn structures. This means

that any interference geometry is revealed fully in the shared fold

profile plane of the two phases, and might be expected to show

Ramsay’s Type 3 refolding, or Type 0 (no refolding), according

to the orientation of the superposed shear folds. Ramsay & Lisle

(2000, pp. 885–901) provided a detailed case study of Type 3

interference caused by two perfectly orthogonal superposed

phases of shear folding (Fig. 2a and b). They reveal the complex

fold patterns and strain histories that result, and discuss some of

the anomalies in small-scale structures and fabrics that might

arise. The most significant effect in Fig. 2b is that the first folds

can be clearly traced, with only small differences locally between

original hinges and final hinges, whereas the second folds are

less persistent, with jumps in axial planes. This model does not

consider the complexities that might arise from differences in

scale of first and second folds, or of major and minor folds and

their vergence.

The use of major and minor folds and of vergence of folds and

cleavage has become an established method in geological map-

ping of folded regions, and the use of vergence in areas of

refolding was discussed by Bell (1981) and Weijermars (1982).

Both suggested that in areas of coaxial refolding, the vergence of

first-phase structures would not be changed by second folding,

and so reversals of vergence (vergence boundaries) could be used

to identify early folds. This was tested by Lisle (1988), who

considered the effects of superimposing a set of second similar

(shear) folds on a bed containing a first cleavage, and he

demonstrated that anomalous ‘cleavage vergence’ and vergence

changes can arise. The study by Lisle (1988) is particularly apt

for our paper, as it appears to have been prompted by specific

structures and vergence features of the Rhoscolyn Anticline, and

examples from here were used to illustrate small-scale vergence

reversals that might lead to misinterpretations of the large-scale

structure. Lisle concluded that vergence may be an unreliable

structural tool in areas of polyphase folding.

The models discussed for refolding, and for the case of coaxial

refolding that we focus on in this paper, have so far involved

superposed similar folds, without any concern for their mech-

anics of origin. However, any discussion of folding or refolding

in rocks with competence contrasts, or of minor and major

folding, cannot ignore the origin of first and second folding; that

is, the buckling mechanics. We thus turn to the question of

whether 2D fold interference effects of the kind shown in Fig. 2b

(after Ramsay & Lisle 2000) would be produced by orthogonal

deformations, in layers with competence contrasts.

Many analogue model studies have considered two-phase

folding in layered materials with rheological contrasts

(Watkinson 1981; Ghosh et al. 1992, 1993; Grujic 1993; Johns

& Mosher 1996), but have been principally concerned with

cross-folding (Type 1 or 2 interference patterns) rather than

coaxial refolding (Type 3). These cited studies all reveal the

importance of the first fold geometry on the development of Type

1 or 2 fold interference in mechanically active two-phase folding.

Ghosh et al. (1992, 1993) classified modes of superposed

buckling, and revealed the 3D complexities that arise in interfer-

1 1

2 2

33

44

a b

c d

Fig. 2. Models for two-phase folding with

parallel fold axes and orthogonal shortening

directions. (a) First phase similar-style folds

with four numbered marker layers

(continuous curves), and F1 axial planes

(dashed lines). (b) Classical Type 3

interference, according to Ramsay’s model

of similar refolding, with F2 folds having

the same geometry as the F1 folds, after

Ramsay & Lisle (2000, fig. 35.6). Same

line symbols as in (a), but also showing

discontinuous F2 axial planes (dotted

traces). (c) Reversal of the F1 folds in (a)

by active unfolding, as a result of ‘equal

and opposite’ D1 and D2 deformations (see

text for details). The layers become straight,

and the original F1 axial planes are folded,

and an associated cleavage (S1) would be

folded on orthogonal F2 axial planes

(dotted). (d) An alternative version of equal

and opposite D1 and D2 deformation, but

where D2 is a passive pure shear

modification of the F1 folds shown in (a).

The folds are opened by homogeneous D2

strain, and F1 axial planes undergo

shortening and folding in a similar fashion

to (c).

THE RHOSCOLYN ANTICLINE UNRAVELLED 119

ing folds of different (major, minor) orders. Johns & Mosher

(1996) examined the effects of varying the competence contrasts.

However, we are unaware of any experimental studies that

examined these variables for coaxial superposed buckling, to

investigate whether Type 3 interference patterns are produced in

layered systems with competence contrasts, or what form they

might have. Our discussion must therefore concentrate on theor-

etical principles.

Let us consider a multilayer comprising competent and

incompetent layers in a first phase of layer-parallel compression,

that produced sinusoidal buckles with 608 limb dips (as in Fig.

2a). If we wish to consider the effects of a second deformation,

equal but orthogonal to the first, for comparison with the similar

folding model (Fig. 2b), we need first to estimate the strain

associated with these first folds. If produced by buckling alone,

this would indicate 35% layer shortening, but it allows for no

internal layer shortening before or during folding. From a review

of analogue and experimental models of single and multilayer

folds (Treagus 1997, figs 19.4 and 19.5), a shortening of 47%

seems a more realistic figure, equivalent to a plane strain with

strain ratio of R ¼ 3:5. The example in Fig. 2 can thus be

considered as two phases of pure shear, each with R ¼ 3:5, the

first (D1) parallel to layering, the second (D2) orthogonally, with

shortening parallel to the axial planes of first folds. (The total

strain is therefore zero.) Instead of assuming that D2 causes a

sinusoidal second displacement pattern, we will consider the

likely response of a mechanically active system.

If the layers behave according to Newtonian rheology, and

there are no changes in layer viscosities, a D2 deformation ‘equal

and opposite’ to D1 would simply reverse the folding instability,

and unfold the layers to their initial straightness (Fig. 2c). Such

perfect reversals of deformation can be produced in analogue

fold models, and in finite-element models with Newtonian flow

laws (S. H. Treagus, unpublished data). Within this ‘unfolded

fold’ (Fig. 2c), a hypothetical first cleavage (assumed parallel to

XY for D1) might become crenulated, to develop an orthogonal

second cleavage. Therefore, although the cumulative effects of

folding and unfolding here lead to zero bulk strain in the system,

this would not be obvious if the two-stage effects were preserved

in fabrics within competent and incompetent layers.

An alternative type of superposed D2 deformation, equal and

opposite in strain value to the first, is shown in Fig. 2d. This is

the result of assuming D2 strain is a homogeneous pure shear of

R ¼ 3:5. The effect is also to open up the first folds of Fig. 2a;

but passively, by changing the limb dip from 608 to 268, to create

a distorted hybrid ‘first fold’. As for the previous model, this

superposed pure shear might shorten and crenulate any first

axial-plane cleavage, to produce an orthogonal second cleavage.

In this model, the polyphase effects would be manifested in

deceptively gentle folds of bedding, but potentially stronger folds

of an earlier axial-plane fabric. This quasi-passive model does

not require that the layers maintain the same rheological

contrasts in D2 as in D1.

It would require quantitative models, and adoption of specific

values for layer viscosities, to reveal the true mechanical behav-

iour of superposed deformation on a particular suite of earlier D1

structures. The modelling would need to be two-stage: first to

generate the first folds; then to deform these again, orthogonally,

under the same model conditions. We are unaware of any

modelling of this kind. From theoretical studies of strain

variations across competence contrasts in layered systems or

related to folding (Treagus 1988, 1993, 1997), we think the finite

deformation effects would be complex. The simplest conceptual

model is to assume quasi-passive D2 deformation, of the kind

shown in Fig. 2d. This may be a suitable approximation to the

bulk response, for some patterns of first folding, and will be

pursued below.

The likelihood of developing new second folds by active

folding or refolding in a multiplayer (rather than unbuckling or

passively distorting the earlier folds in either of the manners

shown in Fig. 2c and d), will depend on satisfying two main

requirements.

(1) Are there sufficient ‘straight’ sections of layering around

first folds that remain in compression for second folds to initiate?

Previous analogue models of buckling of layers in oblique

shortening (Beech 1969; Treagus 1972) showed that significant

folds were produced only in layers initially ,258 to principal

shortening. The orientation does not affect the wavelength and

folds should initiate symmetrically (Treagus 1973), but an S or Z

asymmetry would subsequently develop according to the sense of

obliquity to the strain. Applying these results to (second) folding

of layers in hinges or limbs of first folds, we see that no part of

the folds in Fig. 2a would be in suitable orientation for second

folding. Where the two deformations are not orthogonal, or the

first folds are asymmetric, second folds might develop preferen-

tially on alternate limbs of first folds. Only where first folds are

nearly isoclinal and straight-limbed, might significant second

folding be seen on both limbs (not necessarily with the same

symmetry; Ramsay 1967, fig. 10.21).

(2) Does the layering (singly or multiply), in any section

around a first fold, fulfil the mechanical requirements for

buckling (Johnson & Fletcher 1994) into new folds? It is difficult

to explain how a new set of buckles could properly form, having

the same wavelengths as the first (to satisfy the mechanics of the

system), when a set of earlier folds already exists. One explana-

tion is to assume there are reductions of viscosity contrasts in the

second deformation that would reduce the dominant wavelengths

for buckling and allow smaller folds to grow, which would also

reduce the buckling response. Alternatively, a half fold wave or

whole waves (a ‘fold pair’) might be localized on first fold limbs,

especially on ‘long limbs’, so that the wavelengths of the second

folds are controlled by the size of the first fold limbs, rather than

by the buckling mechanics.

From this reasoning, we conclude that development of new

second folds, or proper coaxial refolding of earlier first folds,

should be the exception in mechanically active systems with

competence contrasts subjected to this type of superposed

deformation. Second folds are likely to be on a smaller scale

than first folds, and systematic fold patterns require first folds

that are straight-limbed and tight to isoclinal. These conclusions

highlight the different interplays of mechanics and geometry: the

manner in which the first folds interfere with, or favour, second

folding, for different orientations of superposed deformation, that

potentially give rise to Types 1, 2 or 3 interference. In rocks with

competence contrasts, we consider that Type 3 interference is

likely to be localized and rare, compared with Types 1 and 2.

This theoretical discussion and its conclusions can now be

applied to real rocks, to help unravel the polyphase structures of

the Rhoscolyn Anticline.

The Rhoscolyn Anticline: review and criticalobservations

Following our brief introduction to the Rhoscolyn Anticline

above (Fig. 1), we now consider it in more detail as a case study

for revealing the superposed effects of two coaxial folding

deformations in layered rocks with competence contrasts.


Review of geology and previous structural interpretations

The dominant structural feature of the Rhoscolyn area of Holy

Island, Anglesey, is the Rhoscolyn Anticline or Antiform (Fig. 1a

and b), illustrated in our map and profile (Fig. 3); detailed

observations are described in the next section. There have been

several conflicting interpretations of the Rhoscolyn Anticline and

its associated minor structures in these Monian Supergroup

(Precambrian or Cambrian) rocks, as simplified in Fig. 1c–f.

Before reviewing these, we summarize the geometry of structural

elements that is common to them all, regardless of their supposed

age or mechanical significance.

(1) Three lithostratigraphic formations are displayed clearly in

an antiformal fold, the Rhoscolyn Anticline (Figs 1 and 3). The

South Stack, Holyhead Quartzite and Rhoscolyn Formations

collectively comprise bedded quartzites, psammites and pelites.

The South Stack and Rhoscolyn Formations consist of alternating

centimetre- to metre-scale psammites, semipelites and pelites,

which sandwich the Holyhead Quartzite Formation, dominantly a

poorly bedded orthoquartzite. These rocks are succeeded by the

New Harbour Formation, dominantly a distinct finely layered

green semipelite.

(2) The major fold plunges to the NE and has an axial surface

that dips to the NW; it has a ‘long’, flatter, limb that dips gently

to the NW, a broad, rounded, hinge zone, and an apparently

‘short’ limb that dips initially steeply to the SE, becoming locally

vertical and overturned, steeply dipping to the NW.

(3) The bedded units on both limbs are affected by intermedi-

ate-scale folds (tens of metres in wavelength) and abundant

minor-scale folds (metres or less in wavelength), most of which

plunge subparallel to, and are congruent with, the major anti-

form.

(4) In the psammitic rocks, most of the minor- and intermedi-

ate-scale folds of bedding referred to above have a penetrative

Fig. 3. (a) Structural map and field data for the Rhoscolyn Anticline, an abbreviated version from our field maps. Readings related to D2 are not shown.

(b) Downplunge projection of (a), based on the average plunge of fold axes of 248/0568 (profile section plane 1468/668 SW). The nine field localities

described in the text are located on map and profile. H shows the major fold hinge trace; X is discussed in the text.


cleavage subparallel to their steep NW-dipping axial surfaces and

to that of the major fold.

(5) The pelites throughout the major fold are dominated by a

crenulation cleavage, which dips at shallow angles to the NW.

They also contain a significant volume of quartz veins, oblique

to bedding, which are folded.

We summarize five previous interpretations of the evolution of

the Rhoscolyn Anticline and its associated minor structures, four

of which are illustrated in schematic form (Fig. 1c–f). Roper

(1992) did not illustrate his interpretation in cross-sectional or

profile view, but we consider it would look very similar to our

own profile (Fig. 3b). We cannot do justice here to the detailed

observations that the previous workers have presented, but have

tried to abstract the essential elements of their work that are

pertinent to their interpretation, and to our reinterpretation.

Readers are referred to these studies for fuller discussion of the

regional geology and tectonics. We omit interpretations (e.g.

Hudson & Stowell 1997) based principally on structures in the

New Harbour Formation, but will address these and their

correlations in another paper.

Shackleton (1969), using sedimentary way-up structures, re-

interpreted the stratigraphic succession and structural interpreta-

tion of the original mapping of the Rhoscolyn area by Greenly

(1919) (Fig. 1c). Shackleton’s four-fold, upward-facing, succes-

sion (Fig. 1d) is accepted by most later workers, and by

ourselves. On the basis of the plentiful minor structures, he

interpreted the Rhoscolyn Anticline as an F1 fold, verging and

facing steeply to the SE. A strong axial-planar penetrative S1

schistosity, fanning around the fold, was observed to be devel-

oped in the quartzites and psammites, parallel to which quartz

veins were segregated in the pelites. Several sets of minor

structures were identified, superimposed upon the Rhoscolyn

Anticline, one of which was said to originate from a vertical

compressional stress field (D2 of this paper).

The view of Shackleton (1969), which we essentially share,

was not challenged until the claim by Cosgrove (1980) that the

Rhoscolyn Anticline was in fact a D2 antiform superimposed on

the flat limb of an earlier major D1 isoclinal fold or kink-band,

facing to the NW (Fig. 1e). This view was based on observations

of the geometry of minor structures (see also Price & Cosgrove

1990, pp. 482–490), which were at variance with those of

Shackleton. Most importantly, an early fabric was observed in

the quartzites and psammites of the South Stack, the Holyhead

Quartzite and the Rhoscolyn Formations, which was folded

around the Rhoscolyn Anticline, although no D1 minor folds

were identified. The well-developed minor folds that verge

towards the major fold were thus identified as D2: In the

quartzites and psammites, a new penetrative S2 cleavage was

developed axial-planar to the minor folds, but fanning around the

major fold; in the steep limb this cleavage was coincident with

S1, but in the hinge zone and flat limb both cleavages were

distinguished. An S2 crenulation cleavage was developed in the

pelites. The geometry of the quartz veins, which were segregated

as planar bodies parallel to S1 in the pelites, was used

particularly by Cosgrove to demonstrate the D2 age of the

Rhoscolyn Anticline and other major folds on Holy Island. His

observed Z-shape geometry of these veins, related to D2 minor

folding in the flat or NW-dipping limbs of the major D2 folds,

was attributed to a top-to-the-NW shear couple set up as a result

of flexural slip between competent beds. The veins on the SE-

dipping limb would have been bodily rotated not folded, and so

the folds with S-shape geometry that were observed in the quartz

veins on the steep limb of the Rhoscolyn Anticline (e.g. Fig. 4b)

were here attributed to a locally developed D3 phase.

Lisle (1988) raised the question of whether the Rhoscolyn

Anticline was a major first or second fold (Fx or Fy in his

terminology), and highlighted its ambiguities, in a discussion of

the nature of structural vergence in refolded regions. As noted in

the preceding section, he showed how a set of shear folds,

superposed on bedding and first cleavage, could produce anom-

alous vergence patterns, and illustrated this with specific exam-

ples from the NW limb of the Rhoscolyn Anticline. Although

presenting evidence that seems more in favour of deducing that

the Rhoscolyn Anticline is a first fold with a related axial-plane

cleavage in the psammites, Lisle concluded that some ambiguity

remained, which could not be solved by vergence information

alone.

Phillips (1991) postulated that the Rhoscolyn Anticline was a

D2 antiform, but for different reasons from those of Cosgrove

(1980). He considered that D1 was responsible for a top-to-the-

SE shearing of the sedimentary pile, producing a bedding-

parallel (S0/S1) fabric in the pelites and in the finer-grained

psammites and a NW–SE chlorite lineation on S0/S1 surfaces;

no folds or fabrics, oblique to bedding, were produced (Fig. 1f).

D2, a progressive continuation of the SE-directed simple shear,

resulted in several major SE-verging folds on Holy Island, such

as the Rhoscolyn Anticline, as well as the dominant minor folds

and their axial-planar fabrics. These fabrics were pervasive in the

psammites and quartzites but contiguous with a crenulation

cleavage developed in lithologies affected by the D1 bedding-

parallel fabric, especially in the pelites. Phillips differed in his

interpretation of the history of the quartz veins from Cosgrove,

in that the veins did not develop parallel to an S1 fabric, but late

in the D1 event as tension gashes in response to the SE-directed

shear; subsequently they were wrapped around the Rhoscolyn

Anticline and affected by the D2 minor folding, thus accounting

for their S-shaped geometry on both limbs. A localized minor D3

folding with an axial-planar fabric (S3) subparallel to S2 is

recognized in the flat limb and hinge zone of the Rhoscolyn

Anticline.

Roper (1992), using evidence that includes fold vergence,

returned to a first-phase (his Dx) origin for the Rhoscolyn

Anticline, interpreting it as a major upright fold associated with

the dominant minor folds and a penetrative cleavage in all

lithologies in the South Stack, Holyhead Quartzite and Rhosco-

lyn Formations. The quartz veins in the pelites were said to be

produced as tension gashes parallel to the first cleavage, as a

result of stress relaxation late in Dx. The second deformation (his

D y) was responsible for creating NW-verging minor folds

(especially of the first cleavage and the quartz veins) with an

axial-planar crenulation cleavage, imposed across the earlier

structure. The second structures were all attributed to a shear

couple dipping NW, which produced inhomogeneous simple

shear zones (likened to Riedel shears) that rotated the original

steep-dipping first cleavage to become the flat limbs of the

second folds. Although this deformation history differs from our

own, the net product in profile view would probably not be very

different from ours, shown below.

These different conclusions for the age of the Rhoscolyn

Anticline and its structures reflect problems in interpretation of

fabrics, and their correlation with minor and major folds, that are

pertinent to many areas of polyphase deformation. In the case of

Rhoscolyn, these distil into questions regarding the correlation of

folds and fabrics in different rock types; whether single pene-

trative fabrics are first, or second, or combined; and what

information can be gained from fold vergence, first or second. It

will become apparent that our interpretation of the development

of the Rhoscolyn area, in subsequent sections, can reconcile

S.H. TREAGUS ET AL .122

many of the differences among the interpretations reviewed

above, and can explain what might appear to be ambiguous

structural relationships.

Description with key localities

We present our interpretation of the Rhoscolyn structures in

terms of a map, downplunge profile and key localities (Fig. 3),

together with structural sketches and photographs (Figs. 4–8).

The profile section was constructed perpendicular to the mean of

all measured first and second fold axes (248/0568). The major

anticline is apparent in both map and cross-section (Fig. 3): an

asymmetric open major fold with a broad and mildly undulating

hinge region. We have located the fold axial or hinge trace at H

in Fig. 3, in common with the above-cited studies. On the scale

of the map and its profile section (Fig. 3), a few substantial

‘minor’ folds can be seen on each limb, with their vergence

supporting the anticline. From details given below, it will become

clear that we regard these and the major anticline as first folds

(F1). There is plentiful way-up evidence in the area (cross-

bedding, graded bedding, load structures) to confirm the upward-

younging succession, and no evidence for major repetitions to

signify earlier isoclinal structures or major thrust repetitions.

Nine field localities are selected that are accessible and

representative, but also chosen to illustrate characteristic hybrid

effects of the two-phase deformation of the area. For each

locality, we describe structures that occur within c. 100 m of

coastal exposures. Our descriptions are fuller for the central part

of the major structure, in the South Stack Formation, as these

observations are critical in our interpretation. Some are compar-

able with Cosgrove’s localities (Cosgrove 1980, fig. 2; Price &

Cosgrove 1990, fig. 18.49), but his cross-section pays scanter

attention to the SE limb, or the NW limb of the Rhoscolyn

Anticline away from the flat hinge–limb region.

Our traverse of the Rhoscolyn Anticline begins on the SE

‘steep’ limb immediately adjacent to Porth y Hwngan (1),

continuing NW on this limb to (2), then via the Holyhead

Quartzite Formation to exposures of the South Stack Formation

near gullies in the hinge zone (3, 4), a well-visited locality on the

‘flat’ limb (5), and towards Porth Gwalch (6), still in the South

Stack Formation. Via Rhoscolyn Head, where the Holyhead

Quartzite Formation reappears, we then examine the Rhoscolyn

Formation on the NW limb at Porth Saint (7) and nearby (8),

ending at Bwa Du (9). (Note that the section of Cosgrove (1980)

stops at locality 7). Our terms will be F1 for folds we interpret as

first folds, F2 for second folds, and S1 and S2 for their related

cleavages. Thus our notation attributes ‘first’ and ‘second’ to the

folding deformations in these rocks, and will differ from other

workers’ numbering schemes based on fabrics or other criteria.

Locality 1 [26657475] is in the Rhoscolyn Formation on the

steep SE limb of the Rhoscolyn Anticline, where cross-bedding

and graded bedding in psammites provide way-up evidence. F1

fold pairs with several-metre wavelengths can be seen with S

asymmetry and steep NW-dipping axial-plane cleavage (see

profile; Fig. 3b), confirming their position on this steep fold limb.

Examples also occur of local F2 folds cross-cutting F1 folds (Fig.

4a), with a shallower-dipping axial plane parallel to crenulation

cleavage (S2) in semipelites to pelites, and spaced S2 locally in

semipsammites.

Locality 2 [26507485] in the Rhoscolyn Formation shows

characteristic structures in pelite beds and the quartz veins they

contain (Fig. 4b), which occur among psammites on this limb of

the fold, where bedding is generally steeply SE dipping. We

interpret the predominant cleavage in psammites as S1, and

observe quartz veins subparallel to S1 in the pelite. The sense of

S1 cleavage refraction from competent psammite to incompetent

pelite supports its position on the SE limb of a major F1 anticline

(Fig. 4b). The F2 folds of quartz veins and of S1 in semipelites

also show S asymmetry (NW vergence), but with significantly

shallow NW-dipping axial planes, parallel to crenulation cleavage

(S2) developed in the pelite, and to localized cleavage that cross-

cuts S1 in some semipsammites.

The major Rhoscolyn Anticline can be mapped from expo-

sures of the Holyhead Quartzite Formation and pelites within it,

and changes of strike and dip are exemplified by exposures

adjacent to the Coastguard lookout [26327520] close to the crest

of the fold. However, the true nature of the major fold hinge, as

it is recognized today, is better seen in the underlying South

Stack Formation of quartzites, psammites and pelites, both in

detail and across gullies to gain larger-scale downplunge views

of the ‘sheet dip’ (10–258 SE) and the mesoscopic fold

vergence.

Locality 3 [26207510] provides spectacular exposures of

approximately symmetrical cylindroidal folds that cascade with a

sheet dip of c. 258 SE, and is close to the major fold hinge

located in earlier cited studies (Fig. 3, H). The folds can be

examined in three dimensions, and fold axes traceable for many

metres are in detail curved or branching. This is a key locality

for recognizing the hybrid nature of the fold structures. We

interpret most of these folds as originally F1, now significantly

modified and distorted by the second deformation, but there are

also examples of F2 folds of bedding and S1 (Fig. 5a). Here, F1

and F2 folds are not always perfectly coaxial, and an angle of up

to 208 between them may locally occur, giving rise to spiralling

S1 –S0 intersections around F2 folds. We do not agree with

Cosgrove’s interpretation that all the mesoscale folds here are F2

folds that fold an earlier cleavage (Cosgrove 1980, figs 5 and 6;

Price & Cosgrove 1990, figs. 18.51 and 18.52), because many of

the folds have axial-planar S1 in their cores (Fig. 5a). The folds

of quartz veins are clear evidence of a significant D2 shortening,

but in a different orientation to the D1 shortening. The combined

effects of F1 –F2 folding leading to the broadly neutral vergence

at this locality reveal important features that we consider critical

in the two-phase history of the Rhoscolyn Anticline. When

unravelling F2 folds of S1 within cleaved semipsammites (Fig.

5a), we find evidence that the F1 cleavage–bedding vergence was

originally NW. A simple ‘undoing’ of the F2 folds in Fig. 5a

suggests that the F1 fold was originally an asymmetric anticline,

with a short thick NW limb and a much thinner and longer SE

limb (now refolded in the F2 folds); that is, an originally NW-

verging asymmetric F1 fold. If representative, these two lines of

evidence suggest that today’s hinge, H, does not mark the

position of the original F1 axis of the Rhoscolyn Anticline.

Locality 4 [26057510] provides further examples of the two-

phase folding effects in the South Stack Formation, and espe-

cially the variations among the different rock types. The minor

fold geometry in psammites is asymmetric, with Z geometry and

SE vergence, but is the combined effect of F1 and F2 folding

(Fig. 4c). Here, F1 folds with distinct hinges and axial-planar to

convergent S1 cleavage are observed adjacent to F2 folds of both

bedding and S1, with flatter axial planes parallel to the crenula-

tion cleavage (S2) in nearby pelites. Other psammite layers

appear only weakly folded, or include straighter regions with

cleavage fans, which suggest some unfolding of F1 folds. After

removal (by eye) of the F2 folding (e.g. in Fig. 4c), the remaining

F1 folds are found to be NW verging, with Z asymmetry.

Semipelite beds reveal chevron folding of S1 (Fig. 4c), whereas

the pelite beds are dominated by crenulation of S1 to produce S2,


and contain strongly folded quartz veins. The effect is that F2

folds of quartz veins and S1 in pelites to semipsammites can

often appear to be tighter than the hybrid F1 –F2 folds of bedding

in psammite to quartzite beds. At this locality there are rare but

important examples of more traditional Type 3 fold interference

in centimetre-thick quartzite beds, where small NW-verging F1

folds and S1 are wrapped around SE-verging F2 folds (Fig. 5b).

This locality also reveals that the minor F1 structures were

initially NW verging, on the original SE limb of the primary

Rhoscolyn Anticline, and that the predominant SE vergence of

the folds is an effect of D2 deformation.

Locality 5 [25957510] exhibits beds with well-developed

slump folding and fluid escape structures (dewatering). Our

structural observations and illustrations (Fig. 4d and e) are taken


Fig. 4. Field photographs of critical

structures at some of the numbered

localities (see Fig. 3), all taken downplunge

(NE). Scales: pencil and notebook are

15 cm long; lens cap is 5 cm in diameter.

(a) F1 minor fold in a psammite bed of the

Rhoscolyn Formation (RF) at locality 1,

showing a steep axial plane and S1 (parallel

to pencil), the NW limb refolded by F2

folds, and S2 cleavage (bottom left). (b)

Pelite bed between two cleaved psammite

beds in the Rhoscolyn Formation at locality

2, showing S-shaped geometry of the F2

folds in quartz veins parallel to S1, and the

anticlockwise sense of S1 refraction from

psammites to pelite. (c) Hybrid F1 and F2

folding (hinges labelled) in a pale psammite

bed in the South Stack Formation (SSF) at

locality 4. Annotations show the changing

orientation of S1 within this folded bed.

The F2 chevron folds of S1 in the semipelite

in the foreground, with axial planes parallel

to S2 crenulation cleavage, should also be

noted. (d) The variable geometry of F1

folds in the South Stack Formation at

locality 5. The uppermost thick psammite

reveals weak folding, whereas the thinner

quartzite layers in the centre are more

obviously folded, with varying geometry

and overall SE vergence. In the foreground,

subvertical quartz veins show variable to

tight F2 folds. (e) Further detail from

locality 5, showing the steep NW-dipping

S1 in psammite (top), continuing

downwards with slight anticlockwise

refraction into the quartz veins in the

underlying semipelite. The variable

geometry of the F2 folds in this array of

quartz veins should be noted. (f)

Relationships of bedding (S0), S1 and S2 in

the South Stack Formation at locality 6.

The lower semipsammite shows kink-like

folding of S1 within the bed, but semipelite

bed above shows stronger F2 chevron

folding of S1, and the development of a

crenulation cleavage (S2).


from cleaved and folded layers stratigraphically between these

sedimentary structures, where the strong asymmetry and SE

vergence of folds would appear to provide support for a position

on the NW ‘flat’ limb of the major Rhoscolyn Anticline. In

detail, however, we consider the asymmetric fold trains seen in

thin psammites to be a combination of F1 and F2 folds (Fig. 4d).

The initially upright F1 folds appear to have been tightened and

rotated into SE vergence by the second deformation, or straigh-

tened out, or have undergone refolding of their limbs, according

to F1 tightness. The final geometry is one of irregular asym-

metric folding on shallow NW-dipping axial planes subparallel to

S2 crenulation cleavage in pelite beds. In contrast, thicker beds

of psammite and semipsammite appear to be only slightly folded

(Fig. 4e), their deformation mainly revealed in the steep axial-

planar S1 cleavage. As for the previous locality, pelite beds

accommodate considerable variations of deformation around the

folds on different scales, and contain a significant volume of

quartz veins of varying continuity, thickness and F2 fold

geometry (Fig. 4e). Any consistent vergence in the folds of

quartz veins is difficult to assess: we find examples of S, M and

Z asymmetry, perhaps reflecting the varying orientations asso-

ciated with S1 cleavage fans around F1 folds, or in localized F2

deformation. We consider the commonest vergence as SE (Z-

shaped) to neutral (M-shaped) (Fig. 4d and e). The subparalle-

lism of quartz veins to S1 cleavage is revealed by a sharp veer

from a variably SE-dipping trend in pelite, to steeply NW-

dipping on approach and entry into psammites (Fig. 4e). This

sense of S1 refraction suggests a position still on the SE limb of

a major F1 anticline. We do not concur with Cosgrove (1980)

that this is evidence that the Rhoscolyn Anticline is a major F2

antiform that refolds F1 structures that are all NW verging.

Instead, we deduce that the original axis of the Rhoscolyn

Anticline is north of this locality, and that the Rhoscolyn

Anticline has been significantly distorted and undergone hinge

migration during an obliquely superposed F2 deformation.

Locality 6 [25807535] is our last site in the South Stack

Formation, still in the broad central part of the Rhoscolyn

Anticline. Successive beds of shallowly NW-dipping psammite

and pelite persist along the cliffs. Psammite beds of .1 m

thickness appear virtually unfolded, but contain a subvertical to

steep NW-dipping S1 cleavage. This is another good locality to

observe the apparent discrepancies in deformation structures in

different rock types and on different scales. Well-cleaved semi-

psammites reveal internal chevron folding of S1 cleavage, on

axial planes at a small angle to bedding (Fig. 4f) or locally

subparallel, and may be accentuated by an S2 cleavage. This type

of refolding feature and vergence effects were described by Lisle

(1988) from this vicinity, and have been discussed above. Some

of the patterns of folded S1 within the beds here are puzzling in

detail, revealing sheaf-like fans (Fig. 5c) that might be preserved

fans related to F1, or F2 geometric effects. These cleaved horizons

show evidence of hybrid folding, as described above, but here

suggestive of overprinting of asymmetric SE-verging F2 folds on

asymmetric SE-verging F1 folds. The like senses of vergence and

asymmetry combine to produce appressed and irregular asym-

metric folds, with lengthened and thinned original F1 long limbs

and a zig-zag shortening of S1 within, and further shortened F1

short limbs (Fig. 5c). Folds of quartz veins in pelite at this

locality are irregular, as before, but we consider the predominant

orientations are steeply NW verging, in S-shaped folds. Taking

the hybrid effects of F1 and F2 folds and the geometry of quartz

veins together, we consider this locality to be on the NW limb of

the initial Rhoscolyn Anticline. Thus, the original axis of the

Rhoscolyn Anticline lies somewhere between localities 5 and 6.

Continuing to traverse northwards, the Holyhead Quartzite

Formation has reappeared (across a fault) around the major

anticlinal closure, and on Rhoscolyn Head (Fig. 3) tight F1 folds

occur in the cliffs [257756], and can be mapped out on the

ground from pelites within the quartzite. The overall vergence is

undisputably SE, on steep NW-dipping axial planes. We then re-

enter the Rhoscolyn Formation, with cliff exposures where

mesoscale folds and fabrics and NW-dipping sheet dips can be

observed from a distance. We interpret these mesoscale folds in

psammite beds to be F1, with strong axial-plane S1 cleavage that

can be clearly observed in local chevron folds (F2) within beds

on F1 long limbs.

Locality 7 is in the bay of Porth Saint [26007585], where three

15 cm quartzite beds in pelite reddened by proximity to faults

reveal a series of F1 folds modified by F2 (Fig. 6). The SE-

Fig. 5. Field sketches of hybrid F1 þ F2 folds and cleavage patterns in

the South Stack Formation, all drawn downplunge (looking NE). Bedding

(S0) and S1 are shown as continuous lines, S2 as dashed lines and quartz

veins in black in (b) and (c). (a) Folded and cleaved psammite layer at

locality 3, the present-day hinge region of the Rhoscolyn Anticline. The

succession of F1 and F2 folds along the layer (revealed by changes of S1

within the bed), which combine to give almost neutral fold vergence,

should be noted. Scale bar represents 1 m. (b) Rare example of F1 folds

refolded by F2 folding and cross-cut by S2, in a thin quartzite bed with

cross-bedding (x) from locality 4. Scale bar represents 10 cm. (c) Detail

of F1 fold pair modified by F2 pair at locality 6, revealed by the F1 axial

traces (dotted lines). In the main psammite layer, S1 is folded (with

rudimentary development of S2) within lengthened flat limbs of F1 folds,

and shows local sheaf-like fans. The hinge region of the syncline appears

much thickened and skewed, as a result of the combination of F1 and F2

folding. Scale bar represents 25 cm.


verging folds reveal convergent cleavage fans in the quartzite

fold cores, good axial-plane penetrative cleavage (S1) in the

pelite in inner arcs, and preservation of arcuate cleavage fans and

finite neutral points (Ramsay 1967, p. 417) in outer arcs. The

effects of F2 are seen in a quartz vein that can be traced from

axial planar in a tight inclined syncline, but is folded around an

adjacent anticline, confirming that is has been tightened during

F2. A crenulation cleavage (S2) is developed in the pelite, axial

planar to the quartz vein folds, and crenulating the arcuate S1

cleavage fan of the original F1 anticline (Fig. 6). These folds

clearly reveal the hybrid effects of F1 þ F2 in mechanically

active layers with competence contrasts, and the change from

SE-verging F1 folds on fairly steep axial planes into tighter and

more strongly inclined hybrid F1 –F2 folds whose axial planes are

subparallel to the regional S2.

Locality 8 [25857595] is a distinct site north of Porth Saint,

where a massive quartzite within the Rhoscolyn Formation crops

out in an open upright F1 syncline (Fig. 7a), marked out by a

thin pelite containing folded quartz veins, within the quartzite.

We see different responses to the second deformation in the thick

competent quartzite and thinner incompetent pelite. In the

quartzite, a dominant rough and anastomosing S1 cleavage

reveals the convergent F1 cleavage fan: it can be followed from

NW dipping through subvertical to SE dipping, around the

syncline (northwestwards) as shown in Fig. 7. Locally, this

anastomosing S1 can appear as a lozenge pattern, but we do not

consider this an S1 and S2 effect (Cosgrove 1980). Where the S1

cleavage in the quartzite is strong, it is folded into angular F2

folds with shallow NW-dipping axial planes, and a rough S2

whose trend can be seen to transect the syncline. F2 folding is

more obvious in the pelite, revealed by ptygmatic folding of

subvertical quartz veins without a consistent sense of asymmetry

or vergence, and an axial-planar S2 crenulation cleavage (Fig. 7b

and c). This is a well-preserved F1 fold that has undergone only

mild second deformation, perhaps associated with slight fold

opening. Its approximately vertical axial plane is considered

close to the original orientation of F1 axial planes.

Locality 9 [26007630], at Bwa Du, sees the last exposures of

the Rhoscolyn Formation around the Rhoscolyn Anticline before

a faulted contact with New Harbour Formation. An asymmetric

SE-verging F1 fold pair with clear axial-plane cleavage (S1) can

be seen in quartzite (Fig. 8), and its cross-bedding provides way-

up evidence, confirming a SE-facing of these SE-verging minor

Fig. 7. (a) Large basin-shaped F1 syncline

in thick quartzite units in the Rhoscolyn

Formation at locality 8, viewed downplunge

(NE), showing locations of three detailed

sketches (b–d) traced from field

photographs. Scale bar represents 15 m. (b)

and (c) show the convergent S1 (continuous

lines) in quartzite (stippled), and veins

following S1 in pelites, where the pervasive

fabric is S2 (dashed lines). Scale bars

represent 40 cm. (d) NW inclined bedding

within quartzites; the lower bed (unshaded)

shows a strong crenulation of S1

(continuous lines) and the development of

S2 (dashed lines) cross-cutting the synclinal

structure. Scale bar represents 5 cm.

Fig. 6. Downplunge view (NE) of series of F1 folds with SE vergence, in

layers of quartzite (stippled) and cleaved reddish semipelite of the

Rhoscolyn Formation (RF), at locality 7. The S1 cleavage fans and

refraction from quartzite to semipelite (continuous lines) are well

preserved, and clearly related to the F1 folds, but the tight inclined F1

fold in the upper-right section appears to have been modified by D2

deformation that folded the quartz veins (black) and created the S2

crenulation cleavage (dashed lines). Scale bar represents 1 m.

Fig. 8. Representation of the significant SE-verging F1 fold pair in the

Rhoscolyn Formation at locality 9 near Bwa Du, based on field

photographs taken approximately downplunge (looking NE). The

quartzite bed (stippled) contains a steep axial-planar S1 cleavage

(continuous lines), and the overlying brown semipsammite (annotated) in

the background reveals the steep S1 and also a shallower NW-dipping S2

cleavage (dashed lines) that cuts across the folds. The central unshaded

region is an effect of perspective. Scale bar represents 1 m.


F1 folds, synthetic with the major Rhoscolyn Anticline. Just

above the quartzite is a brown semipsammite that exhibits two

cleavages (Fig. 8): an S1 cleavage fanning about the axial plane

of the fold pair; and a shallower NW-dipping S2, which clearly

cross-cuts the limbs of the fold pair. We find no evidence in

these rocks of any flat-lying NW-facing F1 structures, as required

in the interpretation by Cosgrove (1980).

Conclusions from field observations

The Rhoscolyn Anticline is a major F1 fold with synthetic

mesoscale folds on each limb, with a sub-penetrative axial-planar

S1 cleavage preserved in psammites, but a predominant crenula-

tion cleavage in pelites (S2). On the small scale, many of the

observable folds are F2 folds of quartz veins and S1 cleavage,

and some folds in thin quartzite beds may also be deduced to be

F2 folds because of inclined axial planes and axial-planar S2.

The ambiguity arises, therefore, as to whether strongly asym-

metric small-scale folds, especially on the NW limb, are F1 or

F2; and if the latter, does this imply that the major structure is an

F2 antiform, as suggested by Cosgrove (1980)? This is not our

conclusion, despite the apparent changes in F2 fold vergence

around the structure.

On detailed investigation, we find that the majority of the folds

that affect bedding in these rocks are original F1 folds that have

been variably modified by the second deformation. According to

the lithology and the initial F1 fold geometry, the effects of F2

are variable. Ambiguous hybrid F1 þ F2 folds occur most in the

open flat hinge region of the Rhoscolyn Anticline (locations 3–

6). At the present-day hinge (locality 3), some F1 structures with

NW vergence appear to be folded around the major structure.

However, on the steep to overturned SE limb (localities 1 and 2),

and the NW-dipping part of the NW limb (localities 7–9),

strongly developed F1 folds have unambiguous vergence in

support of a major F1 fold, and the shallower cross-cutting nature

of the second deformation rules out their interpretation as second

folds synthetic to a major F2 antiform.

Evidence from folded S1 in psammites, especially where the

F2 axial planes and S2 are subparallel to bedding (location 6),

indicates that F2 folding arises from a shortening in a direction

of c. 708 SE in the profile plane (perpendicular to the average S2

trace, 208 NW). This steeply inclined shortening is also deduced

from the range of attitudes of folded quartz veins in pelites

around the major structure. In the next section, we will examine

in more detail the possible effects of a steeply inclined second

shortening on earlier F1 structures, and attempt to remove this

second deformation to reveal the geometry of original F1

structures. However, several lines of field evidence suggest that

the major and minor F1 folds might have been closer to upright

and symmetrical, originally.

We described various structural criteria to suggest that H in

Fig. 3 (including locality 3) is not the major axial trace of the

original Rhoscolyn Anticline. We deduce that this axis was

between localities 5 and 6, placed tentatively at X in Fig. 3, on

the basis of mapped changes in S1 and bedding vergence, and

geometry of folded quartz veins. The c. 260 m distance from X

to H therefore marks the apparent hinge migration of the

Rhoscolyn Anticline, caused by the distortion effected by the

second deformation. This may be considered a quasi-passive pure

shear of the kind shown in Fig. 2d, but here oblique to the F1

axial plane. Thus, the Rhoscolyn Anticline is an F1 anticline, but

has undergone significant modification in its shape and small-

scale structures. It is probably now a more open anticline than

originally, and with a significant section of hybrid structures

where F2 has flattened out, modified or refolded F1 structures on

the original SE limb near the hinge, to make them now part of

the apparent NW ‘flat’ limb of the present-day major structure. If

this region were the main focus of study, and evidence from

further away on the fold limbs disregarded, the major structure

might be deduced to be an F2 fold (antiform).

The overall SE vergence and inclination of the axial planes is

largely a result of the second deformation. We find no field

structural criteria to lead us to interpret the mechanics of either

the first or the second deformation in terms of large-scale simple

shear, in any direction. Instead, the evidence leads us to a simple

model of two phases of pure shear with coaxial intermediate

strain directions parallel to 248/0568 (the average plunge of F1

and F2 folds), and about 708 difference in orientations of

principal strain directions (X and Z) in the profile plane for the

two phases.

Modelling the Rhoscolyn Anticline

Constraining the D1 þ D2 model

The preceding section summarizes the field structural data that

we consider point to a simple two-phase deformation history for

the Rhoscolyn area. These data provide precise constraints for

both deformations, as detailed below, especially in determining

the amount and orientation of the D2 strain. This can then be

removed to reveal the true nature of the D1 structures. For each

deformation, we use X > Y> Z as nomenclature for the principal

axes of strain and their stretch values.

Orientations of principal axes. As F1 and F2 axes are virtually

coaxial (248/0568), we take this as the intermediate axis of strain,

and take Y ’ 1 for both deformations, unless contradictory

evidence emerges. The profile plane of 1468/668SW is thus

common for both deformations, and the two-phase history can be

viewed two-dimensionally, in this plane. The S1 trace is assumed

to reflect the XY plane for D1, but is now modified by D2. The S2

crenulation cleavage is taken as the XY plane of D2 strain. Its

average trace on the profile plane determines the X direction to

be 208 NW (Fig. 9a).

D2 deformation style. The consistency of S2 crenulation

cleavage and F2 folds across the Rhoscolyn Anticline, the

systematic behaviour of specific lithologies, and the absence of

any consistent evidence of any regional shear sense affecting all

the rocks, lead us to reject a model of D2 as simple shear (see

Cosgrove 1980; Phillips 1991; Roper 1992). We consider the

evidence consistent with bulk pure shear (or approximately) for

the D2 deformation; that is, a plane strain pure shear with 208

NW extension (X) and 708 SE shortening (Z) in the profile plane

(Fig. 9). This strain quasi-passively distorted the earlier F1

structures, in the manner of coaxial superposed deformation

discussed above (Fig. 2d), except that the two deformations are

not orthogonal.

Determining the D2 strain from F2 folds. The main constraints

on the D2 strain ratio (X=Z ¼ R), come from the range of

orientations of F2 folds of quartz veins and new F2 folds of

bedding, and their respective shortening values. Buckled quartz

veins in pelites and semipelites across many parts of the

Rhoscolyn Anticline are subparallel to S1, broadly axial-planar

with F1 folds and oriented 60 � 58 NW in profile section (Fig.

3b). Folded veins with S asymmetry have typical shortening

values of c. 20%. Veins that are subvertical to steeply SE dipping

(related to S1 fanning) are more tightly folded into more

symmetrical shapes, with typical shortening values up to 43%. If


this is assumed to be measuring the maximum shortening (Z), it

indicates a pure shear of R ’ 3 affecting these rocks.

Determining the D2 strain from orientations of layers with F2

folds. Now let us consider the information from orientations of

veins or layers containing F2 folds. In the D2 pure shear with

strain ratio R, traces at angle j to shortening (Z) change to j9 by

tanj9 ¼ R tanj (1)

(Ramsay 1967, p. 67). For pure shear of R ¼ 2–4, finite short-

ening sufficient to produce measurable folding (20%) requires

j # 208, and so the orientation range for F2 buckling can be set

at 0 � 208 to Z (Fig. 9a). Oriented with respect to Rhoscolyn D2,

this means veins or layers inclined 90-508 SE in profile view (l–

n, Fig. 9a). Testing different R values in Eq. (1), we find that the

vertical orientation (l) deforms to 628 NW, for R ¼ 3, and would

have folds of S asymmetry (l9, Fig. 9b). This agrees with

observations for S1-parallel quartz veins noted in (3). The other

limit (n, 508SE) deforms to 228 SE and folds with Z asymmetry

(n9, Fig. 9b). The direction of greatest folding (m, 708SE) does

not rotate, and would fold with M symmetry. These asymmetry

and vergence relationships are broadly consistent with the

geometry of folded beds and quartz veins at Rhoscolyn, support-

ing our adoption of R ’ 3 for the superposed D2 quasi-passive

pure shear.

Vergence of F2 folds. The vergence of F2 folds of bedding, or

of quartz veins in pelites, or affecting S1 in semipelites to

psammites, can be compared with the relationships noted above

(Fig. 9). These provide additional constraints on the model for

D2 deformation. We have found the geometry and vergence of

folded quartz veins to indicate D2 pure shear with R ¼ 3. The

more variable vein fold geometry seen in pelites in the central

part of the Rhoscolyn Anticline, where shallow SE-dipping veins

with Z asymmetry are locally seen, can be explained in terms of

veins and S1 originally dipping SE, reflecting a divergent fan into

the major hinge zone, or variations around minor F1 fold hinges.

Our observations do not confirm those by Cosgrove (1980) that

quartz veins in pelites have a predominant SE dip with Z folds

throughout the whole of the NW limb of the Rhoscolyn

Anticline. In psammites, the geometry of F2 folds of S1 reveals

the superposition of D2 pure shear on fanning to axial-planar S1

associated with the Rhoscolyn Anticline. The vergence of F2

folds of bedding is more difficult to assess, because of the hybrid

effects of F1 and F2, particularly in the flat hinge region of the

Rhoscolyn Anticline. The dominant SE vergence of minor folds

on the ‘flat’ limb may reflect true F2 vergence for this orientation

of folds of bedding, but also the distortional effects of D2 strain

on F1 folds that produce hybrid F1 þ F2 folds.

Original F1 fold geometry. Today’s Rhoscolyn Anticline has a

SE-leaning geometry (Fig. 3b), but it cannot be assumed that this

reflects the original F1 fold geometry. ‘Removal’ of a D2 strain

of R ¼ 3, as deduced above, and oriented as in Fig. 9, leads to a

conclusion that the original Rhoscolyn Anticline must have been

less obviously SE inclined and asymmetric than it is now. We

concluded that the average S1 and its parallel quartz veins (now

folded and c. 608 NW) must have been approximately vertical

before D2. This leads us to deduce that the Rhoscolyn Anticline

was an upright fold with vertical axial plane, and perhaps

symmetrical. To constrain the original limb dips of the Rhosco-

lyn Anticline, we must consider the evidence from D2 structures

on the NW and SE limbs of the Rhoscolyn Anticline. The ‘steep’

SE limb of the Rhoscolyn Anticline and its F1 minor folds have

been locally folded (refolded) in F2 folds, and so this limb must

fall in the buckling range shown in Fig. 9. Initial dips of .708

SE would steepen and rotate clockwise towards X, whereas dips

of ,708 SE would decrease dip, so not become a ‘steep’ limb.

As a symmetrical fold with both limbs dipping .708 is a very

tight major F1 fold, we take a conservative lowest limit of 708

limb dip. This means that the SE limb (at its steepest) was

subparallel to principal D2 shortening, Z, so underwent no overall

dip change or steepening of this limb during D2. Thus, any

steepening or overturning of this limb is attributed to D2

shortening that tightened or distorted mesoscale F1 folds and

their long limbs. On the other hand, the 708 NW limb will have a

very different history during D2 strain of R ¼ 3, undergoing a

significant clockwise rotation to a shallower dip of 428 NW (Eq.

1), consistent with the constructed fold profile (Fig. 3b). This

limb was thus oriented for D2 extension, potentially to become a

longer ‘flatter’ limb with bed thinning. The evidence from S1,

which is markedly folded within straight psammite beds, together

with observations of possibly unfolded or significantly skewed F1

folds all confirm that this NW limb has indeed been lengthened

by D2. Therefore all this evidence points to an initial upright

symmetrical F1 Rhoscolyn Anticline, with 708 apparent limb dips

(408 interlimb angle), and a vertical axial plane and S1.

Quantitative D2 model

To arrive at a quantitative model, we tested the distortional

effects of a D2 pure shear with small variations of R about 3.0,

variations in its orientation, and a variety of initial fold shapes

and limb dips. We also compared the results for R ¼ 3 distortion

of symmetric anticlines with 708 limb dips, in fold shapes that

Fig. 9. Orientation of D2 pure shearing in the profile section: (a) original

circle before D2, showing the orientations for extension (X) and

shortening (Z); (b) the D2 strain ellipse, again showing X and Z, where X

is aligned parallel to regional S2. The shaded region shows the limit for

F2 buckling (see text for discussion), and the different vergence geometry

that would be expected in layers parallel to l, m and n are shown in (b)

(l9, m9, n9).


were a sine wave, a circular arc and a parabola (Hudleston

1973). These permutations provide quantitative testing of the fold

shape and D2 strain values that best satisfy all the preceding

properties and constraints for modelling structures at Rhoscolyn.

The deformed shape that is closest to the geometry of the profile

section of the present-day Rhoscolyn Anticline (Fig. 3b) is

obtained from pure shear with R ¼ 3, passively deforming a 708

parabola.

The original parabolic shape of the schematic Rhoscolyn

Anticline is shown in Fig. 10. A parabola with maximum limb

dip, Æ, is expressed in Cartesian coordinates (Hudleston 1973) as

y ¼ tanÆ(�x� x2)=�: (2)

For Æ ¼ 708, the expression becomes

y ¼ 0:875(�x� x2) (3)

as constructed in Fig. 10. A superposed pure shear (R ¼ 3; X

oriented 208 anticlockwise of horizontal) can be written as a

transformation of the (x, y) parabola coordinates to new coordi-

nates (x9, y9) (see Means 1990, eq. 8), which for this deformation

are found to be

x9 ¼ 1:6xþ 0:37y; y9 ¼ 0:37xþ 0:71y: (4)

This is the deformed fold shape (x9, y9) constructed in Fig. 10;

its geometric similarity to our constructed profile of the Rhosco-

lyn Anticline (Fig. 3b) should be noted.

Results

Comparison of the original and deformed fold, aided by the

positions and spacings of graph symbols in Fig. 10, illustrates

the following features of this hybrid antiform, which we relate to

the broad geometry and structures of the Rhoscolyn Anticline.

(1) The deformed fold is more open than the original shape,

strongly asymmetric and SE verging. It has a broader hinge

region, without a distinct axis or hinge point. The original

horizontal is now c. 138 NW.

(2) The fold axis (A) and axial plane (A–P) have deformed to

occupy positions that appear off-centre (A9–P9), not where the

current axial plane or fold hinge might be constructed for the

distorted fold. The current hinge is taken to be where bedding is

perpendicular to S1 (modified) in this deformed fold (Fig. 10,

H), as is the case in the interpretations of the Rhoscolyn

Anticline that we cited above (also Fig. 3, H).

(3) The distance A9–H is thus the effective hinge migration,

approximately halfway round the ‘right limb’ (SE limb). The

broad crest of the distorted fold that is left (NW) of hinge, H,

largely comprises a section of the original right limb (A9–H),

and is thus a hybrid zone of particular significance.

(4) The original right limb (A–C) has become shorter

altogether, but unevenly. Zone A9–H has actually extended,

whereas H–C9 (the current right limb) has shortened by c. 40%,

because its steepest part is subparallel to Z. Thus the current right

limb is a significant short limb for two reasons: it represents only

two-thirds of the original SE limb, and it includes the part of the

original fold that experienced the maximum D2 shortening.

(5) All of the left limb, both its original section (B9–A9) and

its apparent current form (B9–H), has become extended. This is

now a significant long limb. However, its true length is unlikely

to be fully exposed, because a horizontal section (C9 leftwards)

would omit the steeper half of this left limb.

(6) The deformed fold viewed in profile cross-section (Fig. 10)

now comprises three distinct zones that are applicable to the

Rhoscolyn Anticline: (a) a steeply dipping short SE limb that has

shortened; (b) a flat central region, where structures on the

original SE limb have been modified and obliquely stretched,

now to become the hinge to NW-limb region of the hybrid

Rhoscolyn Anticline structure; (c) a long NW limb that has

extended, and its dip decreased.

(7) The original axial plane (A–P) has changed from vertical

to c. 628 NW, and shortened by 20%. Axial-plane structures, such

as cleavage (S1) and S1-parallel quartz veins, would likewise

rotate, and would develop F2 folds with S asymmetry and NW

vergence (compare lines l and l9, Fig. 9).

(8) Any potential F2 folding or refolding of bedding is

restricted to the orientation range for buckling for this D2

deformation (l–n, Fig. 9a); that is, dips of 90–508 SE, in the

vicinity of H–C9. Dips of .708 SE could develop F2 folds with

S asymmetry and NW vergence (l9, Fig. 9b) (as for S1 and quartz

veins). Dips of 50–708 SE could develop F2 folds with Z

asymmetry and SE vergence (n9, Fig. 9b).

This model, with the features listed above, provides a good

simulation of the large-scale geometry of the Rhoscolyn Anti-

cline, as shown in its profile section (Fig. 3b) and the field

descriptions. In the next section, we will build on the simple

model of one fold surface in Fig. 10, to include features to

simulate detailed geological features and mesoscale to small-

scale structures of the Rhoscolyn Anticline.

Fig. 10. Distortion of a parabolic fold with

708 limb dip (x, y coordinates, j) by D2

pure shear with extension X inclined at 208

and strain ratio R ¼ 3, according to the

coordinate transformation in Eq. (4). The

fold becomes the asymmetric curve (x9, y9

coordinates, half-filled squares) with its

inflexion surface (B9C9) dipping 148 to the

left (NW), and the original vertical axial

plane (AP) now A9P9, dipping 628 to the left

(NW). H marks the position that might be

deduced to be the hinge or axis of the

distorted fold, where the fold tangent

surface is perpendicular to A9P9. (H) is its

undeformed position.


Developing the model to unravel two-phase structuresof the Rhoscolyn Anticline

The three main formations in the Rhoscolyn Anticline (Fig. 3)

comprise quartzites, psammites and pelites (and intermediaries)

in various combinations and bed thicknesses. These give rise to

folds on a variety of scales, and the vergence of minor–major

folds has played an important part in previous interpretations of

the area. The presence of buckle folding on many scales also

demonstrates the importance of the competence contrasts in these

rocks. Our approach to modelling these features is not to build

an exact replica of the Rhoscolyn Anticline stratigraphy to

produce the large-scale fold structures shown in Fig. 3b; these

are broadly simulated already in Fig. 10. Instead, we present a

model that includes important key elements to simulate structural

variations that can be related to the Rhoscolyn Anticline on a

number of different scales. We consider three components in Fig.

11: (I) a main bed with bulk properties, to simulate pelite to

semipsammite, with axial-planar S1, and some S1-parallel quartz

veins (v); (II) an upper competent bed, to simulate quartzite or

psammite, with a schematic convergent S1 cleavage fan, mod-

elled on fans in sandstone beds (Gray 1981; Treagus 1982); (III)

a thin layer containing schematic F1 minor folds with changing

vergence and asymmetry, based on patterns given by Ramberg

(1964).

We have not included an incompetent layer with a significant

divergent cleavage fan around the Rhoscolyn Anticline, because

field evidence from the pelites and quartz veins suggests a (more

or less) regional axial-planar S1, not a strongly divergent fan on

each major limb. (Compare, for example, locations 2 and 8, Figs.

4b and 7). Furthermore, theoretical modelling of strain partition-

ing in layers and folds (Treagus 1993, 1997) suggests that in

competent–incompetent alternations, the dominant strain parti-

tioning arises by reduced strains in the competent horizons,

rather than a greatly intensified strain in the incompetent

horizons. Thus, the average layer (Fig. 11, I) provides a reason-

Fig. 11. Schematic model of the distortion

of an upright F1 anticline (a), by D2 pure

shear deformation (b), according to the

model in Fig. 10, but including geological

features relevant to the Rhoscolyn

Anticline. Unit I represents an average layer

or the whole structure, in an F1 similar fold.

After D2 deformation, the S1 cleavage and

veins become inclined (628 NW), and

would develop F2 folds with S asymmetry,

with axial plane S2 crenulation cleavage

(208 NW). The original position of the axial

plane is not a recognizable geometric

feature, and the hinge and axial plane of the

distorted fold would be placed at H–H

(crossed lines). Unit II is a schematic

competent layer with convergent S1

cleavage fan. F2 folding of S1 in this layer

would vary around the main fold, not

occurring on the SE limb, where potential

S2 is at a small angle to S1 in the unit. On

the NW limb, the fanning S1 gives rise to a

steady variation in F2 fold geometry, with

important change in vergence at the

position asterisked. Unit III shows a thin

layer with schematic minor F1 folds (a),

and their distortion and changed vergence

when subjected to passive D2 pure shear

deformation (b). The decrease in

asymmetry towards approximate neutral

vergence on the original SE limb, and an

increase in asymmetry on the NW limb,

should be noted. All these distorted F1 folds

are cross-cut by S2. The position of the

original fold axis (A9) would not be

deduced from the fold vergence in (b).


able approximation for incompetent pelite beds around the

Rhoscolyn Anticline, except where pelite beds are involved in F1

minor folds. Here, heterogeneous deformation and significant

variations in S1 orientation would arise locally, but have not been

specifically modelled.

All these components are subjected to the same model of

homogeneous D2 deformation as shown in Figs. 9 and 10,

although we relax the homogeneity on the small scale to allow

S1 to crenulate and to develop S2, and for S1-parallel quartz

veins to buckle into F2 folds. We do not attempt to model

mechanically active D2 behaviour of the lithological layering in

Fig. 11: this will be addressed below (see Fig. 12). The purpose

of Fig. 11 is to reveal the geometrical effects that might occur

from entirely passive D2 deformation of rocks with different

orientations of S1, and containing F1 minor folds with different

initial asymmetry. We will discuss these features in detail in the

next section.

Modelling the competent bed with the same amount of D2

strain, applied homogeneously as for the whole structure,

provides only a rough approximation. Competent beds with

convergent S1 cleavage fans reflect their ‘stiffer’ layer behaviour

in D1, and a similarly reduced strain might therefore be expected

during D2. These variations cannot be quantified without adopt-

ing specific values for viscosity ratios among the different rock

layers (e.g. Treagus 1988), and so are not attempted here.

Qualitatively, we might expect the principal axes of D2 strain to

refract across competence contrasts. We have not recorded a

significant degree of S2 refraction from incompetent to compe-

tent beds at Rhoscolyn. However, the development of S2 in

quartzites and purer psammites has been noted to be generally

weaker and more localized. Thick quartzites, particularly, might

experience a weaker D2 deformation than the rocks as a whole,

and this has implications for the distorting effects of D2 shown

in our model. For example, the whole of the Rhoscolyn Anticline

might have distorted from an initial upright form to its present

asymmetric SE vergent geometry, as modelled (Figs. 10 and 11),

whereas the Holyhead Quartzite Formation and some of the thick

quartzites in the Rhoscolyn Formation might have preserved

more of its original upright F1 structure and undergone a lesser

degree of hinge migration. There is local evidence for this at

locality 8 (Fig. 7), where an approximately upright open F1 fold

in quartzite reveals a convergent S1 fan, and appears less

distorted by D2 than other mesoscale F1 folds around the

Rhoscolyn Anticline (see Figs. 6 and 8).

The features modelled in Fig. 11 are discussed below, by

structural topic, and presented in terms of the structures of the

Rhoscolyn Anticline and field localities described above. We will

add to this quasi-passive D2 model the question of mechanically

Fig. 12. The development of hybrid F1 þ F2 folds. (a) An asymmetric NW-verging minor F1 fold in a schematic competent layer, such as might be

developed on the SE limb (at 408 dip) of the original anticline (see Fig. 11 for key). S1 and schematic veins (v) are generally vertical, but refract

convergently into the competent layer. (b) F2 folding of the F1 long limb, to achieve the required amount of D2 shortening and ‘passive’ rotation of the 408

SE sheet dip to 108 SE. (See text for further explanation of the model.) In this manner, folds can be produced that change, serially, from F1 with axial-

plane S1 and quartz veins, to F2 folds that fold bedding, S1 and veins, and with axial planes close to the regional S2 orientation. (Compare this model with

examples in Figs. 4c and 5a).


active buckling, and the development of F2 folds of bedding, and

refolding. This involves developing a model for producing hybrid

F1 þ F2 folds (see Fig. 12), which are some of the most puzzling

structures at Rhoscolyn, and have been open to different previous

interpretations. We can then reveal a possible reason for the

observed changes in F1 and F2 fold vergence in the vicinity of

the present-day hinge of the Rhoscolyn Anticline.

Bedding and cleavage, folded fabrics and quartz veins

The average layer (Fig. 11, I) reveals that the original vertical

axial-plane cleavage (S1) has deformed to 628 NW, close to the

average S1 trace in the profile of the Rhoscolyn Anticline (Fig.

3b). The axial-planar S1 remains visible in some semipelites,

cross-cut by a shallower NW-dipping S2 (Fig. 8). In most pelites,

the dominant foliation is the S2 crenulation cleavage, with

significantly shallower 208 NW trace. However, at locality 7, the

S1 cleavage in pelite around F1 folds in thin quartzites is clearly

preserved (Fig. 6).

Quartz veins subparallel to the axial-planar S1 generally reveal

F2 folds of S asymmetry and NW vergence (Fig. 11b), with S2

axial planar, as shown at locality 2 (Fig. 4b). However, we

observe many local variations in the geometry of these quartz

veins and their folds, particularly in the South Stack Formation

in the central part of the Rhoscolyn Anticline (localities 5 and

6), as described above. This may in part be due to irregular

initial vein geometry, but we consider it mainly reflects veins that

track S1 fans around F1 folds on different scales and are thus in a

variety of orientations for F2 folding. In our strain model (Fig.

11), layers or veins originally inclined 90–708 SE should fold

with S asymmetry, whereas those at ,708 SE should fold with Z

asymmetry. Such variations in vein fold geometry are revealed at

locality 5 (Fig. 4d and e).

The S1 fabric of the Rhoscolyn Anticline is generally best

preserved in quartzites, psammites and semipsammites, but some

of these last rocks reveal a shallower-dipping cross-cutting S2

(e.g. locality 9, Fig. 8). The model in Fig. 11b shows that in

theory, S1 would be folded in a different geometry, in different

rock types, according to its original attitude. Our field observa-

tions reveal this to be the case. It is important to note that our

model and our structural observations show S2 cross-cutting both

limbs of the Rhoscolyn Anticline, and cannot be explained as an

axial-planar structure. The geometry of S2 cross-cutting F1 folds

is seen on different scales, at many localities around the

Rhoscolyn Anticline (e.g. localities 1, 8 and 9) (Figs. 4a, 7

and 8).

Let us consider now the D2 effects on the convergent S1

cleavage in the schematic competent bed (Fig. 11, II). The F2

folding effects are shown to vary around the Rhoscolyn Anticline

(Fig. 11b), and the same geometric variations would be seen

around F1 folds on smaller scales, too. On the SE limb, S1 is

extended and rotated to a shallower NW dip, but remains steeper

than S2, and so the oblique cross-cutting effects of S1 and S2

would be clear. The sense of S1 refraction from axial planar

clockwise into competent beds will preserve the F1 NW vergence

on this limb, and this refraction sense is seen at field localities 1

and 2 (e.g. Fig. 4b). Entering the central hybrid zone of the

Rhoscolyn Anticline (Fig. 11b, A9–H), which our model con-

siders is part of the original SE limb, this same NW-verging

sense of the convergent S1 fan in the competent bed should be

preserved, although diminished by the D2 deformation. This is

the opposite sense of refraction into a competent bed from what

would be expected on this part of a major fold, if it were a

single-phase structure with its hinge at H. Our descriptions of

localities 3–5 (e.g. Fig. 4e) noted just such a sense of S1

refraction, and it was these observations, together with vergence

and S1 relationships in minor folds and the geometry of folded

quartz veins, that led us to conclude that the original axis of the

Rhoscolyn Anticline lay between localities 5 and 6 (Fig. 3, X),

shown as A9 in the model (Figs. 10 and 11). This ‘wrong’ sense

of S1 refraction is probably the main reason for Cosgrove (1980)

to deduce that the Rhoscolyn Anticline was an antiform that

refolded earlier bedding–cleavage relationships, but our own

observations clarify that this relationship is peculiar to this

central hybrid region, rather than persistent on the whole of the

NW limb of the Rhoscolyn Anticline.

Our model reveals very different effects of D2 on S1 in the

competent bed (II) on the original NW limb (Fig. 11b). Although

bedding lengthens, the convergent S1 cleavage is oriented close

to the D2 shortening direction, and should undergo significant F2

folding with axial planes (S2) at a small angle to bedding. This

feature of folded S1 within non-folded beds produces the

abnormal cleavage vergence relationships described by Lisle

(1988) from this part of the Rhoscolyn Anticline, and discussed

above. Chevron-style F2 folds of this kind, sometimes of just one

wavelength within a particular psammite bed, are notable

features in flat-lying rocks on this NW limb of the Rhoscolyn

Anticline, as shown at locality 6 (Fig. 4f).

According to our model, there is a critical orientation in the

convergent S1 fan, with original inclination of 708 SE (Fig. 11,

asterisk) where maximum folding of S1 occurs, to create

symmetrical M-shaped F2 folds. This point marks a ‘structural

divide’ on the deforming F1 fold (whether major or minor),

between two regions: (1) where S1 rotates clockwise during D2,

and develops F2 chevron folds with S asymmetry: (2) where S1

rotates anticlockwise, developing F2 folds with Z asymmetry. At

this divide, an internal fanning of S1 is geometrically created

within the lengthening competent bed, unrelated to F1 or F2

folding, as illustrated (Fig. 11b). We regard this as the explana-

tion for the sheaf-like fanning shown on a fold limb at locality 6

(Fig. 5c).

Distortion of F1 minor folds, and their vergence

We consider now a suite of schematic minor F1 folds and their

passive distortion by the D2 deformation, assuming they undergo

no mechanically active F2 refolding (Fig. 11, III). The original

NW-verging F1 folds on the steeper part of the SE limb (Fig. 10,

H–C9) will undergo body rotation and some fold tightening. The

original short limbs become flatter and extended, and original

long limbs are steepened to overturned, and shortened. The total

effect is a decrease in S asymmetry while retaining the NW

vergence. In contrast, on the lengthened NW limb of the

Rhoscolyn Anticline (Fig. 10, B9–A9), these minor F1 folds

would increase their Z asymmetry and their SE vergence.

The most interesting and potentially ambiguous effects are in

the hybrid hinge zone (Fig. 10, A9–H). Here, D2 passively

distorts NW-verging F1 minor folds towards a more neutral

vergence and near-symmetrical (M) forms (Fig. 11b), but the

finite effect will depend on the original tightness and asymmetry

of the F1 folds compared with the D2 distortion. Tighter F1 folds

could retain their NW vergence and original hinges, whereas

more open F1 folds could become significantly modified to

become SE verging and Z shaped, with hinge migration. At

Rhoscolyn, the average F1 fold geometry shows an overall

neutral vergence at locality 3, which is close to the present hinge

(Fig. 3, H), as suggested by Roper (1992) and others. According

to our model, this is not the original hinge of the Rhoscolyn


Anticline, but is a product of the D2 distortion of the original

fold. We consider that the original axis or hinge of the Rhoscolyn

Anticline (see Fig. 11, A), where F1 folds would have changed

their vergence before the second deformation, is somewhere

between localities 5 and 6 (Fig. 3, X). This position is no longer

uniquely revealed by a changing F1 fold vergence, but was

deduced on the basis of other field evidence, such as S1 relation-

ships, as described above.

F2 folding and hybrid F1 þ F2 folds

Observations at localities 3 and 4 reveal that not all the features

can be explained by the passive D2 deformation model shown in

Fig. 11. Some of the features of the minor folds and their fabrics

clearly arise from hybrid F1 þ F2 folding. We thus need to

consider mechanically active behaviour during D2; that is, F2

folding and/or refolding of F1 folds.

Competent and incompetent layers that behaved in a mechani-

cally active way in the first deformation, to develop F1 minor

folds on different scales, would in theory be expected to behave

in a mechanically active manner during D2, and develop new F2

folds. However, we noted in our discussion above of polyphase

deformation and coaxial refolding that mechanically active F2

folding, which either produces new folds in bedding or refolds

F1 folds, is likely to be a complex process in rocks with

competence contrasts that already contain structures on a range

of scales. We can consider two types of mechanical response of

the Rhoscolyn rocks to D2 deformation.

(1) The superposed deformation might simply reactivate the F1

folds, rather than create new F2 folds (see Fig. 2c). Thus, F1

minor folds on the steep SE limb of the Rhoscolyn Anticline,

undergoing shortening in D2, might be reactivated to become

tighter. The lengthening NW limb of the Rhoscolyn Anticline

would undergo the opposite effect: a potential reversal of

buckling that would open up F1 minor folds. There is some

evidence of opened F1 folds at Rhoscolyn, but this can also arise

with the passive deformation model described above, and so

‘active opening’ is rather difficult to prove.

(2) The superposed deformation might initiate F2 folds in

layers that are favourably oriented for buckling, as discussed

already for S1-parallel quartz veins (see also Results, (7)). This

refolding response is the more usual model for polyphase

deformation in the literature, as reviewed above. We have already

discussed F2 folds that affect quartz veins and S1 cleavage in

different orientations. These provide good evidence for the two-

phase deformation history at Rhoscolyn, and allowed us to

quantify the D2 strain in our model. Here, we are interested in

assessing the degree of F2 folding of sedimentary beds, creating

new folds or refolding of F1 folds.

To produce active F2 folds of bedding, layer(s) must be in a

suitable orientation for folding (Fig. 9), be long and straight

enough to accommodate a half or whole fold wavelength, and

satisfy buckling mechanics. Assuming these conditions are all

met, F2 folds would be expected to have different symmetry and

vergence according to bedding orientation around the major fold,

or on subsidiary folds. Steep beds (.708 SE) would develop F2

folds with S asymmetry and NW vergence, as for the axial-plane

F2 structures (Fig. 9b). Such second folds might arise on long

limbs of F1 folds on the SE limb, as described at locality 1 (Fig.

4a). Symmetrical M-shaped F2 folds (neutral vergence) would

theoretically develop on limbs with 708 SE dips. More moder-

ately SE-dipping beds (50–708 SE) would develop F2 folds with

Z asymmetry and SE vergence. These could also arise on long

limbs of F1 folds on the shallower part of the SE limb, or

perhaps on short limbs of F1 folds on the NW limb.

We consider that the most favourable condition for F2 folding

is on long limbs of asymmetric F1 folds on the SE limb. These

are in the D2 shortening direction, and also provide the greatest

layer length to accommodate an F2 fold pair, albeit one that must

have a shorter wavelength than the F1 fold wavelength. This

situation might be further favoured by irregularities in F1 folds

and their wavelengths, so that an F1 fold with an exceptionally

long wavelength could provide an unusually long limb that could

possibly nucleate an F2 fold close to its ideal buckling wave-

length. As noted above, the resulting F2 folds could be NW or

SE verging, or neutral, according to this long-limb orientation.

However, the discussion so far reveals that many geometrical

conditions need to be met for F2 folding, reinforcing our field

observations that F2 folds of bedding are localized structures, not

a systematic refolding of F1 folds.

We illustrate this process in Fig. 12, using a schematic initially

asymmetric F1 fold pair with NW vergence, limb dips of 658

NW and SE, and overall ‘sheet dip’ of 408 SE. Such a fold might

have been seen in the A–H region of the original Rhoscolyn

Anticline (Figs. 10 and 11), which became the hybrid zone after

D2. We assume that the layer is competent, and folded almost

entirely by buckling in both D1 and D2, to retain the original bed

length. In Fig. 12b, the ‘sheet dip’ has rotated to 108 SE, as for

the passive model, but in this case it is accompanied by ‘active’

F2 folding. An F2 fold pair is constructed on the F1 long limb,

with Z asymmetry and SE vergence to meet the D2 strain

requirements, and with geometry that best satisfies other space

requirements. In deriving this construction, we find that the

interlimb bisector (axial plane) of the F2 buckles cannot be

exactly parallel to S2, and this angular discordance controls the

precise orientations of F2 fold limbs and the angle of the long

limbs to S2 (always a small angle). The solution in Fig. 12b best

meets all the geometric criteria, but was also guided by field

observations of cleavage and bedding orientations in hybrid

folds.

The final product is a series of folds appearing to have Z

asymmetry and SE vergence, with longer ‘flat’ limbs, but in

detail they have the form anticline, synform, antiform, syncline,

as shown (Fig. 12b). This type of hybrid folding has been

described from localities 3 and 4, and the example from locality

4 (Fig. 4c) shows notably similar features to those modelled in

Fig. 12. Most important is the sequence from a fold hinge with

an axial-plane S1 (F1), to the next fold that folds S1 together with

bedding (F2) (e.g. Figs. 4c and 5a). Another diagnostic feature of

the model is the small angle between the long limbs of these F2

folds and the crenulation cleavage (S2), also confirmed by field

observations. In Fig. 12b, we also attempt to reproduce some of

the complexities that could arise in quartz veins that are parallel

to S1, in different parts of the hybrid F1 þ F2 folds. Such changes

from veins axial planar to some folds, and yet folded around

others, are all observed in association with hybrid folds at

localities 3–5.

Similar constructions of F2 folds have been made for other

orientations and degrees of asymmetry of F1 folds. Those with

‘sheet dips’ of 308 SE and 508 SE range (deformed range of 08 to

238 SE, the latter being close to H in Figs. 10 and 11) reveal

similar geometric properties of ‘refolding’, and their structural

implications, to those shown in Fig. 12 for the intermediate

orientation. These two other examples (not illustrated) produce

new F2 folds and hybrid F1 þ F2 folds that both suggest overall

SE vergence. The hybrid fold asymmetry is greatest at the

shallower dip, comparable with the present crest or ‘flat limb’ of


the Rhoscolyn Anticline (see locality 5), and least for the model

closest to today’s hinge, H (locality 3), where we produced

almost neutral effective fold vergence in the model.

Active F2 folding is more difficult to explain and model on the

NW limb of the original Rhoscolyn Anticline. Here, the layer

orientations favourable for F2 shortening with possible buckling

are the short limbs of F1 folds. We conclude from field

observations that most of the observed minor folds on this NW

limb are F1 folds that have been modified by D2, quasi-passively,

as described above (Fig. 11). For example, the tight folds at

locality 7 (Fig. 6) appear to have been overprinted by S2, and a

folded quartz vein suggests tightening of one of the F1 folds, but

no significant new F2 folds are recognized. Likewise, the fold

pair at Bwa Du (locality 9, Fig. 8) reflects only the D2

deformation in the superposition of S2 fabrics. These observa-

tions tend to confirm that F2 folding (refolding) of bedding is

restricted, and is probably significant only in the present-day

‘hinge region’ of the Rhoscolyn Anticline. Even there, it rarely

causes coaxial refolding in the traditional sense, with Type 3

cross-sectional patterns (e.g. Fig. 5b), but instead usually gives

rise to compound F1 þ F2 folds (e.g. Figs. 4c and 5a).

The enigma of F1 and F2 vergence changes

Referring back to the mapped structures at Rhoscolyn (Fig. 3),

Cosgrove (1980) considered that F2 folds changed their vergence

at H, supporting his conclusion that the Rhoscolyn Anticline was

an F2 fold. However, Roper (1992) specifically considered F1 and

F2 minor fold vergence, and concluded that there was an F1

vergence change at c. H, leading him to conclude that the

Rhoscolyn Anticline was an F1 structure with its axis at H. Can

both be correct? According to our model, H is not a significant

point on the original Rhoscolyn Anticline, and not where we

would expect to see a change in minor F1 fold vergence around

the major fold. Also, according to our model, H has no structural

significance in D2 deformation, except that it produces the

illusion that there is a major fold hinge here.

We have shown that the combined effect of quasi-passive D2

distortion on mildly NW-verging F1 folds in this region of the

Rhoscolyn Anticline would be to neutralize the F1 fold vergence,

creating an apparent F1 vergence change near H. At the same

time, the change from F2 folds with S asymmetry that formed on

steep F1 long limbs, to folds with Z asymmetry that formed on

less steep F1 long limbs (discussed above), could also be close to

H. Neither of these vergence changes is likely to be a precise

position on the Rhoscolyn Anticline, given (a) the likely

variation in fold geometry of F1 folds, which affects the

neutralizing effect of the D2 deformation on their NW vergence,

and (b) the dependence of F2 fold vergence on whether folding

occurs in beds or F1 limbs with initial dips above or below 708

SE. Thus, today’s hinge (H in Fig. 3) may mark an approximate

‘vergence divide’ for F1 and F2 folds, which could explain why

the Rhoscolyn structures have remained in contention. We

conclude that H is neither the true axis of the Rhoscolyn

Anticline nor a major F2 fold axis.

Calculating the total deformation at Rhoscolyn

From the structures and modelling of the Rhoscolyn Anticline,

we deduce two superposed deformations, with a common inter-

mediate strain axis (Y) subparallel to F1 and F2 fold axes. The

combined effects of these can therefore be analysed in two

dimensions, in the fold profile plane. Our modelling has not

specifically addressed 3D orientations, or reasons for the plunge

of the folds (c. 248), and we have not extended our analyses here

to structures of similar kinds in nearby localities of Monian

Supergroup. Nevertheless, our field studies in South Stack to

Rhoscolyn Formation rocks in other parts of Holy Island support

the deduction that the F1 folds were initially rather tight upright

structures, consistent with the model presented here for the

Rhoscolyn Anticline.

If the total deformation is modelled as two superposed plane-

strain pure shears with a common XZ plane (the profile plane), it

is straightforward to calculate the total finite strain ellipse. From

the symmetrical upright shape of the parabolic F1 fold (Fig. 10a)

in section, the D1 deformation can be deduced to be a pure shear

with horizontal shortening (Z) and vertical extension (X). We do

not have any direct measurement of the D1 strain. A 708

parabolic anticline produced entirely by buckling could be

achieved by a horizontal stretch of 0.58 (horizontal distance/

curve length), but this excludes any layer shortening before or

during the folding, as evidenced by S1 fabrics in all the rocks at

Rhoscolyn. From a recent review of strain associated with

folding (Treagus 1997), we estimate a bulk shortening of c. 0.5

for a multilayer fold of this shape. For plane strain, this would be

a D1 strain ellipse with axial ratio of R ¼ 4.

If this D1 strain ellipse is subjected to the D2 strain of R ¼ 3

deduced above (according to the coordinate transformation; Eq.

4), the resulting finite (D1 þ D2) strain ellipse is found to have

an axial ratio of R ¼ 3:05, with X oriented 558 NW and

Z ¼ 358SE. This total strain is therefore almost the same, in

value, as the D2 strain, but with axes oriented approximately

midway between the D1 and D2 axes. The deformed X direction

for D1 (e.g. a mean S1 trace) ends up only 78 anticlockwise of

the finite X direction. Thus, if a finite D1 þ D2 grain-shape fabric

were to form, it would cross-cut the deformed S1 fabric by only

a small angle, with a slightly shallower NW dip, and might

appear to be axial planar to the deformed fold. However, this

‘total fabric’ would have a significant angle to S2.

Conclusions

(1) Structural observations and modelling demonstrate that the

Rhoscolyn Anticline was an originally upright F1 fold of

parabolic shape with up to 708 limb dip. The second deformation

(D2) has created its present-day asymmetric, SE-facing, more

open form. We estimate that the original anticline axis was c.

260 m NW of the present-day hinge, and that the region between

the old and new hinge is a hybrid zone. Here, original NW-

verging F1 minor folds have been distorted by the D2 deforma-

tion and modified or refolded by F2 folds, to create many of the

ambiguous structures that have given rise to different previous

interpretations.

(2) The structures at Rhoscolyn allow us to quantify the

D1 þ D2 deformation in this part of Anglesey. The original

Rhoscolyn Anticline is consistent with a bulk horizontal short-

ening of 0.5 (R ¼ 4) for the sequence during D1. The D2

deformation is deduced to be a 708 SE-inclined pure shear

shortening with R ¼ 3. This leads to a total deformation of

R ¼ 3:05, with the shortening inclined 358 SE in fold profile

view.

(3) Few studies have addressed the mechanics of superposed

deformation of layered systems with competence contrasts,

where the fold axes are subparallel but the shortening directions

are at a high angle. We provide theoretical and geometrical

arguments why the Type 3 fold interference patterns of Ramsay

(1962) would be unlikely to develop in these rocks, except

locally. Instead, the superposed deformation causes modification


and reactivation of first folds, and creates a variety of hybrid

structures and two-phase fabrics.

(4) This study endorses the conclusion by Lisle (1988) that

fold vergence boundaries are an unreliable method of locating

early folds, in regions of polyphase deformation. The vergence

of minor folds around the Rhoscolyn Anticline has been used in

previous studies, to deduce both that it is a major first fold and

that it is a major second fold. We consider neither is the case,

and the neutral vergence in the present hinge region of the

Rhoscolyn Anticline is the result of the superposition of an

oblique D2 deformation on NW-verging F1 folds.

(5) An important aspect of unravelling the polyphase struc-

tures at Rhoscolyn concerns the relative importance attached to

structures and fabrics in different lithologies. One reason for the

different interpretations is the significance attached to the

deformation and fabrics in psammites (S1), versus those in

pelites (S2) and the quartz veins they contain. We conclude that

least deformed lithologies (quartzites and psammites) provide

clearer clues to the region’s deformation history than those in the

more deformed pelitic lithologies.

We dedicate this paper to the memory of Robert M. Shackleton (died 3

May 2001) and Dennis S. Wood (died 27 April 2001), who introduced

J.E.T. to the geology of Anglesey, and who at Rhoscolyn provided an

enthusiastic introduction to field structural geology for so many geolo-

gists and their students. Thanks go also to the many colleagues and

students, on countless field trips that have helped us form this interpreta-

tion. We appreciate the constructive review of J. Gale, and thank R.

Hartley for drafting some figures. S.H.T. acknowledges a NERC Senior

Research Fellowship, which allowed this paper to be completed.

References

Barber, A.J. & Max, M.D. 1979. A new look at the Mona Complex (Anglesey,

North Wales). Journal of the Geological Society, London, 136, 407–432.

Beech, S.H. 1969. Buckling of single and multi-layers oblique to principal

compressive stress, and its bearing on the problem of asymmetric fold

formation. MSc thesis Imperial College, University of London.

Bell, A.M. 1981. Vergence: an evaluation. Journal of Structural Geology, 3,

197–202.

Cosgrove, J.W. 1980. The tectonic implications of some small-scale structures in

the Mona Complex of Holy Island, North Wales. Journal of Structural

Geology, 2, 383–396.

Ghosh, S.K., Mandal, N., Khan, D. & Deb, S.K. 1992. Modes of superposed

buckling in single layers controlled by initial tightness of early folds. Journal

of Structural Geology, 14, 381–394.

Ghosh, S.K., Mandal, N., Sengupta, S., Deb, S.K. & Khan, D. 1993.

Superposed buckling in multilayers. Journal of Structural Geology, 15,

95–111.

Gray, D.R. 1981. Cleavage–fold relationships and their implications for transected

folds: an example from southwest Virginia, U.S.A. Journal of Structural

Geology, 3, 265–277.

Greenly, E. 1919. The Geology of Anglesey. Memoir of the Geological Survey, UK.

Grujic, D. 1993. The influence of initial fold geometry on Type 1 and Type 2

interference patterns: an experimental approach. Journal of Structural

Geology, 15, 293–307.

Hudleston, P.J. 1973. Fold morphology and some geometrical implications of

theories of fold development. Tectonophysics, 16, 1–46.

Hudson, N.F. & Stowell, J.F.W. 1997. On the deformation sequence in the New

Harbour group of Holy Island, Anglesey, North Wales. Geological Journal,

23, 211–220.

Johns, M.K. & Mosher, S. 1996. Physical models of regional fold superposition:

the role of competence contrast. Journal of Structural Geology, 18, 475–492.

Johnson, A.M. & Fletcher, R.C. 1994. Folding of Viscous Layers. Columbia

University Press, New York.

Lisle, R.J. 1988. Anomalous vergence patterns in the Rhoscolyn Anticline,

Anglesey: implications for structural analysis of refold regions. Geological

Journal, 23, 211–220.

Means, W.D. 1990. Kinematics, stress, deformation and material behavior. Journal

of Structural Geology, 12, 953–971.

Passchier, C.W. & Trouw, R.A.J. 1996. Microtectonics. Springer, Berlin.

Phillips, E. 1991. Progressive deformation of the South Stack and Hew Harbour

Group, Holy Island, western Anglesey, North Wales. Journal of the

Geological Society, London, 148, 1091–1100.

Price, N.J. & Cosgrove, J.W. 1990. Analysis of Geological Structures. Cambridge

University Press, Cambridge.

Ramberg, H. 1964. Selective buckling of composite layers with contrasted

rheological properties, a theory for simultaneous formation of several orders

of folds. Tectonophysics, 1, 307–341.

Ramsay, J.G. 1962. Interference patterns produced by the superposition of folds of

similar type. Journal of Geology, 70, 466–481.

Ramsay, J.G. 1967. Folding and Fracturing of Rocks. McGraw–Hill, New York.

Ramsay, J.G. & Huber, M. 1987. The Techniques of Modern Structural Geology.

Volume 2. Folds and Fractures. Academic Press, London.

Ramsay, J.G. & Lisle, R.J. 2000. The Techniques of Modern Structural Geology.

Volume 3. Applications of Continuum Mechanics in Structural Geology.

Academic Press, London.

Roper, H. 1992. Superposed structures in the Mona Complex at Rhoscolyn, Ynys

Gybi, North Wales. Geological Magazine, 129, 475–490.

Shackleton, R.M. 1969. The Precambrian of North Wales. In: Wood, A. (ed.)

The Pre-Cambrian and Lower Palaeozoic Rocks of Wales. University of

Wales Press, Town, 1–22.

Thiessen, R.L. & Means, W.D. 1980. Classification of fold interference patterns: a

reexamination. Journal of Structural Geology, 2, 311–316.

Treagus, S.H. 1972. Processes in fold development. PhD thesis, University of

Manchester.

Treagus, S.H. 1973. Buckling stability of a viscous single-layer system oblique to

the principal compression. Tectonophysics, 19, 271–289.

Treagus, S.H. 1982. A new isogon–cleavage classification and its application to

natural and model fold studies. Geological Journal, 17, 49–64.

Treagus, S.H. 1988. Strain refraction in layered systems. Journal of Structural

Geology, 10, 517–527.

Treagus, S.H. 1993. Flow variations in power-law multilayers: implications for

competence contrasts in rocks. Journal of Structural Geology, 15, 423–434.

Treagus, S.H. 1997. Modelling deformation partitioning in folds. In: Sengupta,

S. (ed.) Evolution of Geological Structures in Micro- to Macro-Scales.

Chapman & Hall, London, 341–372.

Watkinson, A.J. 1981. Patterns of fold interference: influence of early fold shapes.

Journal of Structural Geology, 3, 19–23.

Watkinson, A.J. & Cobbold, P.R. 1981. Axial directions of folds in rocks with

linear/planar fabrics. Journal of Structural Geology, 3, 211–217.

Weijermars, R. 1982. Vergence boundaries: an extension of the vergence concept.

Journal of Structural Geology, 4, 407–409.

Received 14 December 2001; revised typescript accepted 24 May 2002.

Scientific editing by Haakon Fossen


Documents

Superposed Deformations and Their Hybrid Effects