Continuous arch wire closing loop design, optimization,

and verification. Part I

Raymond E. Siatkowski, DMD
Akaroa, New Zealand

In Part I, a systematic approach to closing loop design for use in continuous arch wires is
presented. The design process uses Castigliano's theorem to derive equations for moment-to-force
ratio (M/F) in terms of loop geometry. The equations are used to optimize designs by optimizing
M/F to produce tooth movement via translation. Further refinements are performed with finite
element simulations of designs. In Part II, predicted results are verified experimentally. The result of
this process is a new design, the Opus loop, which is capable of delivering a nonvarying target M/F
within the range of 8.0 to 9.1 mm inherently, without adding residual moments via twist or bends
(commonly gable bends) anywhere in the arch wire or loop before insertion. The resulting precise
force systems delivered with nonvarying M/F can move groups of teeth more accurately to achieve
predetermined anteroposterior treatment goals for esthetics and/or stability. In Part II the
experimental results show that the loops must be bent accurately to achieve their design potential.
The negative impact on M/F of various dimensional changes to the loop design are presented.
Experimental data are presented illustrating the improved performance of the new design over
standard available designs. Suggested applications of the design for varying anchorage
requirements are presented, along with a case report in which rigorous protraction requirements
were met. (Am J Orthod Dentofac Orthop 1997;112:393-402.)

Translation of a free body occurs when a
net applied force has a line of action that passes
through the body's center of mass. A constrained
body translates when the force's line of action passes
through its center of resistance, a fixed point deter-
mined by the geometry of the body and its con-
straints. It is often desired to move malposed teeth
via translation to avoid localized areas of high stress
and strain that can produce traumatic resorption of
tooth roots and investing structures. Tipping and
root movement produce localized high stress areas.
Because of anatomic limitations in the oral
cavity, it is not usually possible to devise an
intraoral mechanism to deliver a force whose line
of action passes through a tooth's center of resis-
tance. Equivalent force systems, comprised of
forces and moments, are instead applied to brack-
ets bonded to the tooth's crown to achieve trans-
latory movement.
Two approaches can be used to apply the force
systems necessary to trigger the biologic phenomena
that result in space-closing movement of individual
teeth or groups of teeth ("en masse"). The first
Supported by a grant from the American Association of Orthodontists
Foundation.
Reprint requests to: Dr. Raymond E. Siatkowski, P.O. Box 118, Akaroa,
New Zealand.
Copyright © 1997 by the American Association of Orthodontists.
0889-5406/97/$5.00 + 0 8/1/77118
approach l involves supplying the appropriate mo-
ments to the teeth via a continuous arch wire that
passes through orthodontic brackets (delivering the
moments via couples, equal and opposite noncol-
linear vertical forces, at the mesial and distal bracket
extremities); the appropriate force is applied via
elastomeric modules or coil springs, for which vari-
ous alloys available. The resulting instantaneous
moment-to-force ratio (M/F) determines the initial
displacement of the tooth (or teeth) within its
viscoelastic constraining periodontal ligament
(PDL). That initial displacement causes a stress-
strain distribution within the PDL, which triggers
the bone remodelling processes that yield the de-
sired direction of tooth movement. As the tooth
moves, the applied force decreases (elastomerics are
further subject to force decay with time). The ap-
plied moment can increase or decrease, dependent
on the arch wire configuration. Therefore, the M/F
changes as the tooth moves, and the tooth responds,
typically progressing from controlled tipping (center
of rotation at the root apex) to translation to root
movement. Such progression may not produce the
most efficient or the least traumatic tooth move-
ment. Wire-bracket friction is a variable factor as
the moving teeth displace along the arch wire with
this approach, making it difficult to accurately pre-
dict M/F.
The second approach involves bending arch wire
393
304 Siatkowski American Journal of Orthodontics and Dentofacial Orthopedics
October 1997

TOOTH Bracket-Center of Inclination to MIF for Translation

Resistance Distance Oeelusal Plane (°) (ram)
(ram)

Maxillary
1 9.6 59.0 8.2
2 8.6 63.0 7.7
3 9.7 78.6 9.4
4 8.6 86.1 8.6
5 8.6 88.8 8.6
6 8.5 83.5 8.4
7 7.6 92.2 7.6
Mandibular
1 8.0 71.0 7.6
2 8.9 71.0 8.4
3 10.3 84.0 10.2
4 8.6 87.8 8.6
5 8.6 84.2 8.6
6 8.5 80.5 8.4
7 7.6 68.6 7.1

MIF FOR
TRANSLATION

21l 12 8.0 8.8 65a 356

8.8 6543 345~
8.6 76543 34567
321 I 123,, 8.4 8.5 65 56
8.2 765 567
21 T"12 8.0 9.1 653 356
9.0 (~543 3456
8.6 76543 34567
3211 123 8.8 8.5 65 56
8.2 765 I 567

Fig. 1. Moment/force ratios (M/F) required to achieve translation for individual teeth and
groups of teeth of mean dimensions assuming no marginal loss. (From Table 2, Siatkowski
RE. Force system analysis of V-bend sliding mechanics. J Clin Orthod 28(9):543, 1994,
with permission.)

loops of various configurations, sectionally (to de-
liver the desired M/F to an individual tooth) or
segmentally or in a continuous arch wire (to deliver
the desired M/F to several teeth). This approach is
friction-free; when activated, the arch wire loops
distort from their original configuration; as the tooth
(or teeth) moves, the loop gradually returns to its
undistorted (preactivated) position, delivering the
energy stored at the time of activation. Brackets are
not sliding along the arch wire during the process.
An essential characteristic of closing loops for or-
thodontic space closure is that they are free of
friction as they act. Groups of teeth can therefore be
moved with more accurately defined force systems
for more precise anchorage control to achieve treat-
ment goals more readily than methods in which
friction plays a role. Presently available designs,
having insufficient inherent M/F, are clinically diffi-
cult to apply with sufficient accuracy to achieve these
goals when placed in a continuous arch.
American Journal of Orthodontics and Dentofacial Orthopedics Siatkowski 395
Volume 112, No. 4

Helix of N half-turns and

o, rod,us

F~ ~ F

K = F = Load
Displacement

E bh
K
+ 4 H R2 + R N ~ [ H2 + - - ]
3 2

Haack, D.C., AJO 49(5]: 330-344, May, 1963

Fig. 2. Load/displacement rate equation for a vertical closing loop, derived by Haack. 17

RELATIVE LOAD - DEFLECTION RATE

VERTICAL LOOP (0.016" x 0.022")

EE 6
E u2..-~

"r" ¢3 ~ RELATIVE INFLUENCE

H>n>>D
(25
o 4
&
< 3

tll
->
I--
2
..1
tu
n- 1

I I I I
O0 4 6 8 10 H, Loop Hr. (mm) e~e
(D = 3.Smm, 1 Helix)
I I I I I I
2 3 4 D, Loop Diam. (mm) " - _ I
(H = 8mm, 1 Helix)
, ,
0 1 No. of Helicies o---o

Fig. 3. Load/displacement (F/D) relative to a standard vertical loop 8 mm high and 3.5 mm
diameter with one helix in 0.016 x 0.022 inch stainless steel wire when loop height, loop
diameter, and number of helices are varied.

Besides configuring the looped arch wire to proach; if F/D is large, stress levels traumatic to the
deliver proper M/F to achieve the direction of tooth PDL, alveolar bone, and roots can be delivered at
movement desired, the loop's load-deflection rate very small loop activations that are difficult to
(F/D) assumes added importance with this ap- deliver precisely. Also, large F/D, requiring small
396 Siatkowski
F.<.J°L> M
\
0 M
M H 2 + H R a:N +
F 2 H + N~R
Fig. 4. Moment-to-force ratio (M/F) equation for a vertical closing loop. Variables the
same as defined in Fig. 2.
activations, deactivate after small tooth movements;
if the M/F is not constant, the PDL stress distribu-
tions change rapidly as the tooth cycles from con-
trolled tipping to translation to root movement.
Most closing loop designs 2,3'4'5 offered to date opti-
mize for low F/D at the expense of M/F. Some 6,7
achieve lowered F/D by using arch wires formed
from alloys with reduced Young's Modulus.
Bowley et al. 8 laid the groundwork for holo-
graphic measurements and finite element analyses
to determine the location of the centers of resis-
tance for individual teeth and thereby, the M/F
necessary to achieve translation. That M/F is simply
the distance between the center of resistance and
the bracket, measured perpendicular to occlusal
plane (the plane of the arch wire). Subsequent
studies by Burstone et al., 9 Dermautet al., 1° Tanne
et al., 11,a2 Pedersenet al., 13 and Andersen et al. a4
found the location of the center of resistance for
individual teeth. On the basis of mean tooth dimen-
sions from Wheeler, 15 these results are summarized
and extrapolated to derive the M/F required to
achieve translation for individual teeth and groups
of teeth in Fig. 1,1 assuming no marginal bone loss.
The derived M/Fs vary from 7.1 to 10.2 mm for
individual teeth and 8.0 to 9.1 mm for groups of
teeth (more gingival placement of brackets will
lower these values). No closing loop design previ-
ously has been capable of delivering M/F at these
levels, most having inherent M/F of 4-5 mm or less.
To achieve net translation, orthodontists have had
to add residual moments to the closing loop arch
wire with angulation bends (gable bends) anterior
and posterior to the loop, a posterior gable bend and
angulations within the loop, or a posterior gable
bend and anterior wire-bracket twist (anterior root
torque). Adding these residual moments has several
disadvantages:
American Journal of Orthodontics and Dentofacial Orthopedics
October 1997
2 R2 0 E b h3
12 F[2 H + N~R)
1. The teeth must cycle through controlled tip-
ping to translation to root movement to
achieve net translation (lower Young's Mod-
ulus materials go through fewer of these cycles
for a given distance of space closure).
2. The correct residual moments are difficult to
achieve precisely in linear materials.
3. The resulting ever-changing PDL stress distri-
butions may not yield the most rapid, least
traumatic method of space closure.
If a closing loop design capable of achieving
inherent, constant M/F of 8.0 to 9.1 mm without
residual moments were available, en masse space
closure with uniform PDL stress distributions could
be achieved. Such a mechanism would be less de-
manding of operator skill to apply clinically and
might provide more rapid tooth movement with less
chance of traumatic side effects. More speculatively,
there is some evidence from animal studies 16 that
intermittent force systems may produce more effi-
cient tooth movement, perhaps 1 hour of force
system application followed by 7 hours of rest. A
mechanism based on micromotor technology could
be designed in the future to activate/deactivate
closing loops on a time schedule, but it would
require loops without residual moments to produce
a true rest period.
The purpose of this study is to systematically
derive and verify a closing loop design capable of
delivering the required M/F inherently, without
adding residual moments, so that more precise force
systems with nonvarying translatory M/F can be
delivered by dosing loops in a continuous arch.
Although derived in 1881 and used since in
engineering design for statically indeterminate sys-
tems, the first application of Castigliano's theorem
in orthodontic spring design was by Haack 17 in 1963.
American Journal of Orthodontics and Dentofacial Orthopedics Siatkowski 397
Volume 112, No. 4

T-LOOP
R• 2L+d "-~R

-d- _J_

[H2+4L (H+R) + 3L2 + 2~RH + 4R 2 d(H+2R+2L) +d21- EI0/F

F
[2H + 4L + 2~R + d] ~b~

I = b.__hh3
12
L-LOOP

[H2+2L (H+c+d) + 2L2+ (d+c)2 + 2xRH + 4R 2] - EI0/F

i =

F
2(H+2L+d +c+~R)

OPUS LOOP I' L "l

•-) 2R

n t
H

[2k2+2L (H+c) + 3L2+ 2~R (2H + k) + 12R2] - EI~)/F

4=
2(2k + 2L + 3~R)

c=H+2R
k=H+R

Fig. 5. M/F equations for T-loops, L-loops, and Opus loops.

Waters 18-2° began applying the method for analyzing
existing designs in the 1980s. After initial consultations
with Haack, the use of this method as an optimizing
tool (maximizing M/F for any given loop configura-
tion) was undertaken. The method is limited by as-
sumptions of small strains (the wire behaves perfectly
elastically) and of small geometric changes as the loop
is activated. It is a useful guide in following mathemat-
ical trends for design optimization, however.
Once a new design has been derived, analyses
using the finite element method (FEM) can be helpful
for further design optimization, because larger dis-
placements can be simulated. After the advent of the
FEM in the 1960s, 21 there were few applications s'22 of
the technique in orthodontics until commercial soft-
ware packages became available. 2,s These generate
distortion plots of the loops after loading as well as
displacement, load, and moment data from which M/F
and F/D are found easily.
True verification of any proposed new design
requires load, displacement, and moment measure-
ments of actual samples of the loop. The verification
process will be presented in Part II of this study.
MATERIAL AND METHODS
Theoretical investigations using Castigliano's theorem
were undertaken for vertical loops, T-loops, and L-loops,
each with increased projected M/F. Detailed perusal of
the mathematical trends suggested a new design, the
398 Siatkowski American Journal of Orthodontics and Dentofacial Orthopedics
October 1997

2.0
VERTICAL L

EE
E ~x
i[ t[
"r¢l~

GS
0

1.0
N
LU

k-
.J
W

4 10 12 14 16
H, L o o p Ht. ( m m ) =
(D = 3 . 5 r a m , 1 H e l i x )
I i I f I m I i D, L o o p Diam.(ram) =--=
1 2 3 4 (H = 8ram, 1 Helix)
I !
No. of H e l i c i e s o - - o
1 2 (H = 8 m m , D = 3 . 5 m m )

Fig. 6. M/F for a vertical loop relative to the standard loop when loop height, loop
diameter, and number of helices are varied.

EE
EE
==-~ 2.0
i[ []

if)

[4.
1.0

LU

ILl
n"
D = 3.5nlm
1 Loop

00 10 15
LOOP HEIGHT (mm)

Fig. 7. M/F for the loops described in Figs. 5 and 6 relative to the standard vertical loop
when loop height is varied.
American Journal of Orthodontics and Dento¢acial Orthopedics Siatkowski 399
Volume 112, No. 4

Tt
H
L -'~1
N ~
~H (L = 10mm
R = 0.5mm

2.0

E
E /'~'~--------------~3-d (H = L = 10mm
L~X
..m
J ~ R = 0.5mm)
II "I-

GE R(H = L= 10mm)
o
i-

I.I. 1.0

ILl
>
m

I--
_1
ILl
n" I I I
0 1 2 d(mm) [] []
I I I I I

0 1.0 2.0 R, mm "

°0 ; .... 1; .... 1;
H, mm (L = 10mm, R = 0.5mm) e------e
L, mm (H = 10mm, R = 0.5ram) o o

Fig. 8. T-loop M/F relative to the standard vertical loop when loop height (H), end radius
(R), loop length (L), and leg separation (d) are varied.

"Opus loop." Specificvertical loops and Opus loops were
then simulated by use of Ansysed 5.0-56 FEM software
(Swanson Analysis Systems,Houston, Pa.), the new design
being further optimized.
RESULTS
T h e o r e t i c a l Findings
Haack's equation for load/displacement ratio for
vertical loops 17 (Fig. 2) was investigated for 0.016 ×
0.022 inch s.s. wire. For an 8 mm high loop with one
apical helix, the 0.016 × 0.022 inch F/D is minimized
with a loop diameter of 3.5 mm (Fig. 3).
In collaboration with Haack, Castigliano's theo-
rem was applied to derive M/F for such vertical
loops (Fig. 4). An outline of the steps to use
Castigliano's theorem follows:
1. The loop is broken up into its most basic
geometric components for individual stored
energy analysis, that is, straight lines and
segments of circles.
400 S i a t k o w s k i American Journal of Orthodontics and Dentofacial Orthopedics
October 1997

EE I J
L ./
EE X

II II Z

o
I--

2.0
n H L/H/L=IO mm\

L 10 m]
LU
>
i.
m

I'-
...I
LU
rr
1.0 L
0
, I , I
1.0 2.0 R(mm)-- --
I , , , , I , , , , I

(~0' ' ' 5 10 15

H, mm(L=10mm, R=O.5mm) :
L, mm (H 10mm, R 0.5 mm) o o

Fig. 9. Opus loop M/F relative to the standard vertical loop when loop height (H), radius
(R), and loop length (L) are varied.

O P U S 70 . 0 1 6 X . 0 2 2 ss, 1 3 m m IBD, 1.5rm~ to h e l i x e n d

Fig. 10. Ansysed FEM distortion plot output for an off-centered 0.016 × 0.022 inch s.s.
Opus 70 loop for 13 mm interbracket distance with an activation force of 75 g. Dotted line:
passive position. Solid line: activated position.
American Journal of Orthodontics and Dentofacial Orthopedics Siatkowski 401
Volume 112, No. 4

lOmm "1
lmm (
T
J;
,jr,o, °
10ram

~'~ ['--Up to 1.5mm

Fig. 11. Dimensions of the standard Opus 70 loop.

2. A force system, in equilibrium, is applied to where x = the net deflection of loop ends in
the loop ends. the horizontal direction and F x = horizontal
3. Equations for the energy stored in each loop component of applied force
component are derived from: , These equations are then solved to find M as
a function of Fx.
(M 2 ds 7. The resulting equation is divided by F x to find
U = J 2 E I for bending
an equation for M/F. The resulting equations
are shown in Figs. 4 and 5 for various loops of
where U = stored energy, M = applied mo- interest. The vertical loop with an apical helix
ment, E = Young's modulus, I = wire cross- is widely used clinically, was the original loop
sectional moment of inertia perpendicular to design analyzed by Haack, 17 and is therefore
the bending direction, and ds = increment of chosen as the reference loop for comparison
length in the direction of the basic geometric with other designs.
component's length.
4. The equation for the total energy stored in the Because wire size and Young's modulus have little
loop by the applied force system is found by effect on inherent M/F (later verified by experimental
summing the energy stored in each loop com- results in Part II), the second term was ignored as a
ponent: simplifying assumption in the calculations to produce
Fig. 6. Fig. 6 predicts that the greatest effect on raising
UTO T = ~ Ux M/F is to increase loop height. Increasing the number
x=l of apical helices has a lesser effect. M/F is maximized
at 3.5 mm when loop diameter is varied.
5. Castigliano's theorem is applied by taking Equations were then independently derived for
three partial differentials of the total stored M/F for T-loops and L-loops using the same theo-
energy equation: rem (Fig. 5). Making the same simplifying assump-
0 UTO T
tion as above by ignoring the last term in the
-~+~ numerator (-EIQ/F) again, found to be valid by later
0M
experiment, the resulting M/F relative to the opti-
where ¢~ = the angular change at one loop end mized vertical loop as a function of loop height with
and [3 = the angular change at the other loop loop dimensions as shown appears in Fig. 7. For
end (~ + [3 = 20 in Fig. 4), equal loop heights, theory predicts increased inher-
ent M/F for a T-loop configuration and even more
0 UTOT
for an L-loop configuration. This trend suggested
0Fy - y
placing a helix somewhere in the apical portion of
where y = the net deflection of loop ends in the "L" to increase M/F further. The position cho-
the vertical direction and Fy = vertical com- sen is practical in bending the loop in a continuous
ponent of applied force, and arch. The equation derived for M/F for this config-
uration, named the Opus loop, is shown in Fig. 5 and
0 UTO T plotted in Fig. 9.
-- X
0 Fx Parameters in the M/F equations of Fig. 5 for the
402 Siatkowski
T-loop and Opus loop were varied and plotted in
Figs. 8 and 9, respectively, again the EIQ/F term is
ignored. Inspection of both figures reveals that M/F
is maximized when as much wire is placed as high as
possible in the loops (H and L large, R small).
Opus loops 10 mm high, 10 mm long, and 0.5
mm radius in 0.016 × 0.022 inch s.s. wire, 0.018 ×
0.025 inch s.s. wire, and 0.017 × 0.025 inch TMA
(Armco Corp., Glendora, Calif.) were simulated via
the Ansysed FEM software, dividing the loop geom-
etry into 65 beam elements with Poisson's ratio of
0.3 and Young's modulus, 30 × 106 psi for s.s. wire
and 10.4 × 106 psi for TMA. The anterior end was
fixed and the posterior end 13 mm distal was
constrained from moving in all but the horizontal
direction, a reasonable simulation of actuality when
closing loops are placed in a continuous arch. When
centered in the interbracket distance, the M/F at the
bracket connected to the helix end of the loop
always exhibited at least three times the M/F of the
other end.
Angulation of the vertical legs was then varied in
5-degree increments until M/F was equal at both
ends. This occurred when the legs were angled at 70
degrees to the plane of the brackets. Simulation runs
were then performed for the original design with
vertical legs (Opus 90) and the legs angulated at 70
degrees (Opus 70) in off-centered locations within
the 13 mm span. The M/F increased as the loop was
positioned closer to one bracket than the other (the
V-bend effectS), more when the helix end was closer
than when the other end was. An activated symmet-
ric closing loop acts as a V-bend located at the
loop's center. This can be visualized by examining
the horizontal legs of the vertical loop in Fig. 4. An
activated asymmetric loop acts as an off-centered
V-bend. An example of a distortion plot output is
shown in Fig. 10. The dimensions of the optimized
design are shown in Fig. 11.
Discussion and Conclusion
The systematic theoretical derivation of the new
closing loop design requires verification. True verifica-
American Journal of Orthodontics and Dentofacial Orthopedics
October 1997
tion requires load, deflection, and moment measure-
ments of actual samples of the loop. Experiments to
provide that information will be presented in Part II,
along with suggested clinical applications and a case
report.
REFERENCES
1. Siatkowski RE. Force system analysis of V-bend sliding mechanics. J Clin Orthod
1994;28:539-546. Addendum: 1995;29:37-8.
2. Haskell BS, Spencer WA, Day M. Auxiliary springs in continuous arch treatment.
Part I. Am J Orthod Dentofac Orthop 1990;98:387-97.
3. Shaw MM, Waters NE. The characteristics of the Ricketts maxillary canine
retractor. Eur J Orthod 1992;14:37-46.
4. Gjessing P. Biornecbanical design and clinical evaluation of a new canine retraction
spring. Am J Orthod 1985;87(5):353-62.
5. Faulkner MG, Fuchshuber P, Haberstock D, Mioduchowski A. A parametric study
of the force/moment systems produced by T-loop retraction springs. J Biomech
1989;22:637-47.
6. Burstone CJ, Goldberg AJ. Beta titanium: a new orthodontic alloy. Am J Orthod
1980;77:121-32.
7. Hilgers JJ, Farzin-Nia F. Adjuncts to bioprogressive therapy: the asymmetrical "T"
arch wire. J Clin Orthod 1992;26:81-6.
8. Bowlcy WW, Burstone CJ, Koenig HA, Siatkowsld RE. Prediction of tooth
displacement using laser holography and finite element technique. In Heron RE,
editor. Biostereometrics 74. Falls Church, Va: Am Soc Photogramrnetry, 1974:241-
73.
9. Burstone CJ, Prypumiewicz R J, Bowley WW. Holographic measurement of tooth
mobility in three dimensions. J Periodontol Res 1978;13:283-94.
10. Derrnaut LR, Kleutghen JPJ, DeClerck HJJ. Experimental determination of the
center of resistance of the upper first molar in a macerated, dry human skull
submitted to horizontal headgear traction. Am J Ortbod 1986;90:29-36.
11. Tanne K, Sakuda M, Burstone CJ. Three-dimensional finite clement analysis for
stress in the periodontal tissue by orthodontic forces. Am J Orthod Dentofac
Orthop 1987;92:499-505.
12. Tanne K, Koenig HA Burstone CJ. Moment to force ratios and the center of
rotation. Am J Orthod Dentofac Ortbop 1988;94:426-31.
13. Pedersen EH, Andersen KL, Melsen B. Tooth displacement analyzed on human
autopsy material by means of a strain gagc technique. Eur J Orthod 1991;13:65-74.
14. Andersen KL, Pedersen EH, Melsen B. Material parameters and stress profiIes
within the periodontal ligament, Am J Orthod Dentofac Orthop 1991;99:427-40.
15. Wheeler RC. A textbook of dental anatomy and physiology, Philadelphia: WB
Saunders, 1965.
16. Gibson JM, King GJ, Keeling SD. Long-term orthodontic tooth movement
response to short-term force in the rat. Angle Orthod 1992;62:211-15.
17. Haack DC. The science of mechanics and its importance to analysis and research
in the field of orthodontics. Am J Orthod 1963;49:330-44.
18. Waters NE. The mechanics of retraction components used in fixed appliance
therapy. In: Huiskes R, vanCarnpen DH, deWijn JR, editors. Biomechanics:
principles and applications. Boston: Martinus Nijhoff, 1982:445-50.
19. Waters NE. The characteristics of asymmetrically disposed orthodontic retraction
loops. In: Perren SM, Schneider E, editors. Biomechanics: current interdisciplinary
research. Boston: Marinus Nijhoff, 1985;283-8.
20. Waters NE. The mechanics of asymmetric retraction loops used in fixed appliance
therapy. J Biomech 1990;23:1093-1102.
21. Zienkiewicz OC. The finite element method in structural and continuum mechan-
ics. London: McGraw Hill, 1967.
22. Yang TY, Baldwin JJ. Analysis of space closing springs in orthodontics. J Biomech
1974;7:21-8.