ORIGINAL ARTICLE

Maxillary arch perimeter prediction using

Ramanujan's equation for the ellipse
David D. Chunga and Richard Wolfgrammb
Brooklyn, NY, and Vacouver, Wash

Introduction: The prediction of arch perimeter gained when expanding the arch or proclining the anterior inci-
sors is particularly important in the decision of tooth extraction vs nonextraction therapy for orthodontic treat-
ment. Methods: Correlation of the measured perimeter of the maxillary dental arch and the calculated
perimeter applying Ramanujan's equation for the perimeter of an ellipse was evaluated with 30 diagnostic casts
of untreated maxillary dental arches. Both linear and circumferential measurements were made directly on the
midbuccal surface of these maxillary arches. Results: Ramanujan's equation had a high level of correlation
when comparing the measured perimeter of the maxillary dental arch and the calculated perimeter with the
0.01 level (1.2% error) using a 2-tailed t test. The results were applied to Ramanujan's equation to predict
maxillary arch perimeter gained by expansion or proclination of the incisors. Conclusions: The ellipse is an ac-
curate geometric model of the maxillary arch form. The average amounts of maxillary arch perimeter gained were
0.73 mm per millimeter of intermolar expansion and 1.66 mm per millimeter of incisor protrusion. (Am J Orthod
Dentofacial Orthop 2015;147:235-41)

A
rch perimeter or circumference prediction is an
essential component when planning comprehen-
sive orthodontic treatment. This is most evident
when seeking to resolve dental crowding or arch-length
discrepancy (ALD). To treat the ALD, an orthodontist
may choose to expand the arch,1 procline the anterior in-
cisors,2 distalize the posterior dentition, reduce teeth in-
terproximally,3 or extract teeth.4 The decision of the
treatment modality must take into account the occlusion,
facial esthetics, and stability of the result.5 Expansion of
the arch and proclining of the anterior incisors can be
an effective way of gaining arch perimeter when such
treatment is indicated.6 However, without accurate pre-
diction of the arch perimeter to be gained when proclining
or expanding the arch in a borderline extraction patient,
the orthodontist may underestimate the space gained
and erroneously prescribe extractions that will then
require excess space closure or, conversely, expect to
gain more space and be unable to resolve the ALD.
a
Program director, Division of Orthodontics and Dentofacial Orthopedics,
Department of Dentistry, Maimonides Medical Center, Brooklyn, NY.
b
Private practice, Vancouver, Wash.
All authors have completed and submitted the ICMJE Form for Disclosure of
Potential Conicts of Interest, and none were reported.
Address correspondence to: David D. Chung, Maimonides Medical Center, 4303
13th Ave, Brooklyn, NY 11219; e-mail, dchung@maimonidesmed.org.
Submitted, March 2014; revised and accepted, October 2014.
0889-5406/$36.00
Copyright 2015 by the American Association of Orthodontists.
http://dx.doi.org/10.1016/j.ajodo.2014.10.022
Ricketts et al7 suggested guidelines that state that
each millimeter of canine expansion provides for a
1-mm increase in arch perimeter, and that 1 mm of
molar expansion increased the perimeter by 0.25 mm.
Adkins et al8 estimated the arch perimeter gained in pa-
tients treated with a hyrax expander using photographs
and pretreatment and posttreatment study casts.
Regression analysis indicated that changes in premolar
width were predictive of changes in arch perimeter
(r2 5 0.69) at approximately 0.7 times the premolar
expansion. Germane et al9 developed a mathematic
model using a spline function to compare quantitatively
the effects of various types of orthodontic expansion on
the mandibular arch perimeter with average arch dimen-
sion, intermolar width, intercanine width, and midline
arch length. The results of that study indicated that
when the canine width and incisor positions were held
constant, an initial 1-mm increase in molar width pro-
duced approximately a 0.27-mm increase in perimeter,
the second millimeter produced an increase of
0.31 mm, and the fth millimeter of molar width in-
crease was related to a perimeter increase of 0.41 mm.
When the incisor positions were xed, each millimeter
of canine expansion provided a 0.73-mm increase in
arch perimeter. When arch perimeter was increased by
incisor advancement, it was nearly 4 times as effective
in increasing arch perimeter as was molar expansion, de-
pending on arch constriction. Hnat et al10 produced a
method for forecasting alterations in arch perimeter
235
236 Chung and Wolfgramm

related to various width increases in each dental arch

based on combined beta and hyperbolic cosine functions
using average values for the initial preexpanded arch
widths and depths of untreated patients. Because the
canine-to-molar width expansion ratios are a function
of the point of application of the expansion force relative
to the center of resistance of the dentomaxillary com-
plex, 3 canine:molar ratios1:1, 1.25:1, and 1.5:1
were examined. Hnat et al found that if the maxillary
molar width is expanded by 6 mm (3 mm per side) and
the canine:molar expansion ratio is 1.25:1, then the
arch perimeter alteration is 15.4 mm. Correspondingly,
the mandibular arch perimeter alteration is 15.6 mm,
when the 2 arches are in an Angle Class I occlusion.
The studies mentioned above represent a range of
values that can be difcult to interpret when attempting
to predict arch perimeter gained by proclining anterior
incisors or expanding the posterior dentition. To accu- Fig 1. Diagram of an ellipse, where a is the semimajor
rately predict the arch perimeter, the resulting arch axis, and b is the semiminor axis. An ellipse was t on
the maxillary dental arch schematically.
form must be understood. Numerous studies have re-
ported that normal dental arches approximate certain
geometric curves. Nevertheless, throughout years of The dental arch can be modeled by an ellipse by as-
investigation, various geometric forms have continually signing the semimajor axis a the perpendicular distance
been used to describe the shape of human dental arches, from the line intersecting the distobuccal cusps of the
including ellipse, parabola, hyperbola, and catenary maxillary rst molars and the facial surfaces of the
curve.1,11,12 Currier13 concluded that the ellipse is the maxillary central incisors, and assigning the semiminor
best geometric gure for describing the form of both axis b the midbuccal surfaces of the distobuccal cusps
the maxillary and mandibular dental arches when land- of maxillary rst molars. The correlation of the dental
marks on the facial (outer) surfaces of the teeth are used arch to a geometric gure not only permits a static rep-
for comparisons. However, most recently, the arch form resentation but also, by adjusting variables, allows a dy-
has been mathematically modeled more complexly using namic representation that aids in arch perimeter
a computer-aided spline and beta function, bringing prediction.
into question the accuracy of the classic elliptical arch The primary objective of this study was to quantita-
form.9,14 tively demonstrate the mathematical correlation of
An ellipse is a plane curve that results from the inter- Ramanujan's equation for the perimeter of an ellipse
section of a cone by a plane in a way that produces a and the maxillary arch perimeter. Our secondary objec-
closed curve. The calculation of the perimeter of an el- tive was to apply the equation to predict the arch perim-
lipse requires an innite series of calculations to be eter gained by expansion of the molars or proclination of
exact. Several mathematicians have attempted the the incisors. The results will aid clinicians in determining
formulation of an adequate approximation. However, the appropriate treatment for borderline extraction
the equation formulated by Srinivasan Ramanujan15 in patients.
1914 is widely considered to be the most accurate. The
approximation requires 2 values: a and b. The semimajor MATERIAL AND METHODS
axis (denoted by a in Fig 1) is half of the major axis: the
Samples were taken from plaster study models of the
line segment from the center, through a focus, and to
maxillary arch. The study included 30 study models of
the edge of the ellipse. Likewise, the semiminor axis (de-
untreated patients with well-aligned dentitions who
noted by b in Fig 1) is half of the minor axis.16
had received no prior orthodontic treatment. Each pa-
tient had a full dentition from rst molar to rst molar.
Perimeter of an ellipse 5

  p Mild spacing was permitted if the dentition was well
pa1b 11 3h 10  4  3h aligned. The sample consisted of the maxillary arches
 of 11 male and 19 female subjects ranging in age from
where h5a  b2 a1b2 11 to 29 years, with an average age of 15 years.

February 2015  Vol 147  Issue 2 American Journal of Orthodontics and Dentofacial Orthopedics
Chung and Wolfgramm 237

Table. Measured vs predicted perimeter of the maxil-

lary arch
U6 to 6 U1 to U6 Calc perim Meas perim Error
Sample (mm) (mm) (mm) (mm) (%)
1 59.7 38.5 107.79 107.4 0.36
2 55.4 36.5 101.32 101.8 0.48
3 59.0 37.4 105.45 103.3 2.08
4 55.5 35.5 99.72 98.7 1.04
5 62.0 34 102.15 100.0 2.15
6 54.6 37.6 102.58 102.0 0.57
7 57.8 34.2 99.29 97.2 2.15
8 54.9 36.7 101.29 101.0 0.29
9 54.6 33.8 96.24 95.7 0.57
Fig 2. Measurement of the arch length (a), intermolar 10 55.9 33.1 96.06 94.2 1.98
width (b), and arch perimeter was made directly on 11 58.9 39.4 108.71 106.1 2.46
each study cast. 12 53.9 36.1 99.56 99.5 0.06
13 57.7 34.8 100.20 98.3 1.93
14 51.0 34.7 95.11 95.0 0.12
Linear measurements were taken from the midbuccal 15 55.1 38.3 104.12 104.3 0.17
16 56.4 37.2 103.21 102.2 0.99
surfaces of the distobuccal cusps of the maxillary rst 17 61.8 36.1 105.40 103.7 1.64
molars with a digital caliper (6-in composite digital 18 58.7 36.5 103.74 102.1 1.61
caliper; Cen-Tech, Camarillo, Calif). The contact point 19 60.4 37.5 106.65 104.9 1.67
of the caliper with the molar was marked with a vertical 20 56.8 36.2 101.84 99.1 2.77
line using a ne lead pencil for subsequent measure- 21 56.9 35.8 101.25 99.5 1.76
22 59.8 39.9 110.20 109.4 0.73
ments. The perpendicular distance was then measured 23 57.8 35 100.60 99.5 1.11
from the line intersecting the distobuccal cusps of the 24 55.0 34.2 97.20 95.7 1.57
maxillary rst molars and the facial surfaces of the 25 52.8 37.6 101.30 100.6 0.69
maxillary central incisors using a straight-edge ruler 26 54.5 37 101.50 102.2 0.68
and the caliper (Fig 2). Measurement of the arch perim- 27 51.3 32.9 92.32 93.4 1.16
28 61.6 37.9 108.20 108.2 0.00
eter was made directly on each model from the distobuc- 29 52.8 35.7 98.09 95.6 2.61
cal cusps of the maxillary rst molars with a light-gauge 30 52.0 33.5 93.83 93.2 0.68
wire (0.012-in stainless steel; American Orthodontics, The predicted perimeter was calculated by entering the linear mea-
Sheboygan, Wis) that contacted the midbuccal surface surements into Ramanujan's equation for the perimeter of an ellipse.
of each tooth (Fig 2). The light-gauge wire was marked The percentage of error of prediction from the measured values was
at the distobuccal cusps of the rst molars and then laid also recorded. U6 to 6, Intermolar width; U1 to U6, arch length
at on grid paper. The marks made on the light-gauge (depth); Calc, calculated; perim, perimeter; Meas, measured.
wire were transferred to the grid paper and measured
with the caliper.
To determine the measurement reproducibility and was the perpendicular distance measured from the line
the method error, 10 models were randomly selected intersecting the distobuccal cusps of the maxillary rst
and remeasured 4 weeks after the initial analysis. The molars and the facial surfaces of the maxillary central in-
reliability was evaluated with Dahlberg's formula.17 cisors. The value for b was the linear measurement taken
The error for the linear measurement from the midbuccal from the midbuccal surfaces of the distobuccal cusps of
surface of the distobuccal cusps of the maxillary rst the maxillary rst molars divided in half.
molars was 0.04 mm. The error for the measurement The value that was calculated with Ramanujan's equa-
of the perpendicular distance from the line intersecting tion for the perimeter of an ellipse was then compared
distobuccal cusps of the maxillary rst molars and the with the value measured directly on each model from
facial surfaces of the maxillary central incisors was the distobuccal cusps of the maxillary rst molars using
0.06 mm. The error for measurement of the arch perim- a light-gauge wire that contacted the midbuccal surface
eter made from the distobuccal cusps of the maxillary of each tooth with the Pearson correlation test.
rst molars using a light-gauge wire that contacted
the midbuccal surface of each tooth was 0.11 mm. RESULTS
The data were then inserted into Ramanujan's equa- The results show correlation to the 0.01 level, with a
tion for the perimeter of an ellipse (Table). The value for a Pearson correlation of 0.97 (Fig 3) and only a 1.2% error

American Journal of Orthodontics and Dentofacial Orthopedics February 2015  Vol 147  Issue 2
238 Chung and Wolfgramm

Fig 3. The measured perimeter of the maxillary arch was plotted against the calculated perimeter.
Pearson correlation 5 0.97. Correlation is signicant to the 0.01 level (2-tailed).

Fig 4. The change in maxillary arch perimeter when altering the intermolar width was compared while
xing the arch length (depth). The variance was not clinically signicant.

when using the equation for the perimeter of an ellipse
to calculate the perimeter of the maxillary arch.
By applying Ramanujan's equation for the perimeter
of an ellipse to a dynamic analysis, the arch perimeter
gained can be calculated. When the arch length or depth
is xed at the average depth of 36 mm, and the trans-
verse change is plotted, 0.73 mm of arch perimeter is
gained per millimeter of expansion. This value increases
to 0.76 mm per millimeter of expansion when the arch
depth is xed at 32 mm and decreases to 0.71 mm per
millimeter of expansion when the arch depth is xed
at 40 mm, giving a range of 0.05 mm depending on
arch depth; this was considered clinically insignicant
(Fig 4). When the transverse dimension is xed at the
average length of 56 mm and the incisor protrusion is
plotted, an average of 1.66 mm of arch perimeter is
gained per millimeter of protrusion. This value increases
to 1.70 mm per millimeter of protrusion when the

February 2015  Vol 147  Issue 2 American Journal of Orthodontics and Dentofacial Orthopedics
Chung and Wolfgramm
Fig 5. The change in maxillary arch perimeter when altering the incisor protrusion was compared while
xing the intermolar width. The variance was not clinically signicant.
transverse length is xed at 50 mm, and 1.63 mm per
millimeter of protrusion when the transverse length is
xed at 62 mm, giving a range of 0.07 mm depending
on the transverse dimension; this was considered clini-
cally insignicant (Fig 5).
DISCUSSION
As demonstrated in this study, Ramanujan's equation
for the perimeter of an ellipse can be adapted to calcu-
late the maxillary arch perimeter with acceptable accu-
racy (1.2% error). It can then be inferred that accurate
predictions of expansion and protrusion of incisors
may be made by adding the expansion distances to a
and b in the formula. When using Ramanujan's formula,
we found that (1) for every millimeter of intermolar
expansion, we can expect to gain an average of
0.73 mm of maxillary arch perimeter; and (2) for every
millimeter of incisor protrusion, we can expect to gain
an average of 1.66 mm of maxillary arch perimeter.
These results correlate well with the study of Adkins
et al8; they found an average increase in arch perimeter
of 4.7 mm for an average molar expansion of 6.5 mm,
which translates to an average of 0.72 mm perimeter
gain for every millimeter of expansion. Germane et al9
reported that when the incisor positions were xed,
each millimeter of canine expansion provided a 0.73-
mm increase in arch perimeter. As they pointed out, in-
dividual isolation of the width changes will cause an
abnormal arch form, and alterations in the canine region
will alter the incisor positions and premolar and molar
American Journal of Orthodontics and Dentofacial Orthopedics
239
widths; this makes comparison of these studies difcult.
Germane et al reported that increasing the midline arch
length by incisor advancement was nearly 4 times as
effective in increasing arch perimeter as was molar
expansion, whereas our results showed just over twice
the arch perimeter gained in incisor proclination over
molar expansion. Hnat et al10 showed that if the maxil-
lary molar width is expanded by 6 mm (3 mm per side),
and the canine:molar expansion ratio is 1.25:1, then the
arch perimeter alteration is 15.4 mm, giving an average
of 0.9 mm of arch perimeter gained per millimeter of
expansion; this is notably higher than the arch perimeter
expected according to this study.
As studied previously by Ricketts et al,7 Germane
et al,9 and Hnat et al,10 change in intercanine width is
an important variable to consider for the stability of
treatment. Although Ramanujan's equation cannot be
used directly to calculate the perimeter change by
altering intercanine width, it can be interpreted indi-
rectly by a graphic method as follows. In addition to
values a and b as shown in Figure 2, a perpendicular dis-
tance from dental midline to the line connecting the
canine cusps is measured from a study cast. After trans-
ferring the measured distance to the computer-
generated half ellipse, the change in intercanine width
is measured from the dimensionally accurate graphs
before and after expansion of intermolar width (Fig 6).
Finally, the change of intercanine width can be related
to the change of arch perimeter calculated from Rama-
nujan's equation. In the particular case shown in
Figure 6, a 2-mm expansion of intermolar width resulted
February 2015  Vol 147  Issue 2
240
Fig 6. A half ellipse was graphed using the measured di-
mensions of semimajor (a) and semiminor (b) axis of sam-
ple number 18. The intermolar width was expanded by
2 mm, and the new arch form was also graphed. After
measuring the perpendicular distance from the facial
midline of the maxillary incisors to the line connecting
the midlabial surfaces of the right and left canines from
the study cast (c), it was transferred to the computer im-
age. The change of intercanine width (d) was measured
directly on the computer image.
in 1.64 mm of expansion of the intercanine width. Since
Ramanujan's equation predicts a 1.47-mm increase of
the arch perimeter for 2 mm of expansion of the inter-
molar width, we can conclude that 0.9 mm of arch
perimeter can be gained from 1 mm of expansion of
the intercanine width. A future study will include the
relationship of the width change to arch perimeter in
the canine, premolar, and second molar regions.
A limitation of this study was that the distobuccal
region of the rst molars was used as a semiminor
axis. A dental arch, especially a tapered arch form, usu-
ally ts more precisely into the ellipse if a second molar
or the farther distal region is used as a semiminor axis.
However, we deliberately used the rst molar area
because many patients who need expansion of their
dental arch have only their rst molars erupted at the
initial clinical examination. Another limitation is that
this study included only the maxillary dental arch. A
study of the dynamic changes of the mandibular arch
perimeter will be our next goal.
An interactive computer program was written for
Matlab 2014 software (MathWorks, Natick, Mass)
(Appendix) to calculate an arch perimeter by asking the
user to input the following: intermolar width (semiminor
axis), arch length (semimajor axis), the change of
February 2015  Vol 147  Issue 2 American Journal of Orthodontics and Dentofacial Orthopedics
Chung and Wolfgramm
intermolar width, and the change of arch length. The
program generates the number of perimeters of an
arch before and after expansion or protrusion as well
as the net change of the perimeters. It also generates
the accompanying arch forms for visual aid. An accurate
estimation of the perimeter change by altering expan-
sion or proclination of the incisors will ultimately aid cli-
nicians to plan for more accurate treatment in borderline
extraction patients.
Guidelines suggest that when treating Class I crowd-
ing or protrusion with less than 4 mm of ALD, extraction
is rarely indicated; with 5 to 9 mm of ALD, nonextraction
or extraction treatment is possible depending on hard-
tissue and soft-tissue considerations; and 10 mm or
more of ALD almost always requires extraction.5 A pop-
ular trend in orthodontics to resolve an ALD is to expand
the arch and procline the anterior teeth. However, it
must be well understood that the limitations of expan-
sion include esthetic and stability considerations, and
that transverse expansion across the mandibular canines
cannot be reliably maintained.18 Excessive transverse
expansion can lead to fenestration of the roots through
the lateral wall of the alveolar process.5
CONCLUSIONS
Ramanujan's equation for the perimeter of an ellipse
can be adapted to calculate the perimeter of the maxil-
lary arch with only 1.2% error. The amounts of arch
perimeter gained were 0.73 mm per millimeter of expan-
sion and 1.66 mm per millimeter of incisor protrusion
with this prediction.
REFERENCES
1. Bishara SE, Staley RV. Maxillary expansion: clinical implications.
Am J Orthod Dentofacial Orthop 1987;91:13-4.
2. Steiner CC. Cephalometrics for you and me. Am J Orthod 1953;39:
729-54.
3. Bolton WA. Disharmony in tooth size and its relation to the anal-
ysis and treatment of malocclusion. Angle Orthod 1958;28:
113-30.
4. English JD. Mosbys orthodontic review. St Louis: Mosby; 2009. p.
132.
5. Proft WR. Contemporary orthodontics. 4th ed. St Louis: C. V.
Mosby; 2007.
6. Mutinelli S, Manfredi M, Cozzani M. Mathematic-geometric model
to calculate variation in mandibular arch form. Eur J Orthod 2000;
22:113-25.
7. Ricketts RM, Roth RH, Chaconis SJ, Schulhof RJ, Engel GA. Ortho-
dontic diagnosis and planning. Denver, Colo: Rocky Mountain Or-
thodontics; 1982. p. 194-200.
8. Adkins MD, Nanda RS, Currier GF. Arch perimeter changes in rapid
palatal expansion. Am J Orthod Dentofacial Orthop 1990;97:
194-9.
9. Germane N, Lindauer SJ, Rubenstein LK, Revere JH Jr, Isaacson RJ.
Increase in arch perimeter due to orthodontic expansion. Am J Or-
thod Dentofacial Orthop 1991;100:421-7.
Chung and Wolfgramm 241

10. Hnat WP, Braun S, Chinhara A, Legan HL. The relationship of arch
length to alterations in dental arch width. Am J Orthod Dentofacial
Orthop 2000;118:184-8.
11. Battagel JM. Individualized catenary curves: their relationship to
arch form and perimeter. Br J Orthod 1996;23:21-8.
12. Jones ML, Richmond S. An assessment of the t of a parabolic curve to
pre- and post-treatment dental arches. Br J Orthod 1989;16:85-93.
13. Currier JH. A computerized geometric analysis of human dental
arch form. Am J Orthod 1969;56:164-79.
14. Braun S, Hnat WP, Fender DE, Legan HL. The form of the human
dental arch. Angle Orthod 1998;68:29-36.
15. Michon GP. Perimeter of an ellipse. In: Final answers. Updated on
February 28, 2014. Available at: http://www.numericana.com. Ac-
cessed May 1, 2014.
16. Prasolov VV, Tikhomirov VM. Geometry. Providence, RI: American
Mathematical Society; 2001. p. 80.
17. Dahlberg G. Statistical methods for medical and biological stu-
dents. New York: Interscience Publications; 1940.
18. Little RM, Wallen TR, Riedel RA. Stability and relapse of mandib-
ular anterior alignment: rst premolar extraction cases treated
by traditional edgewise orthodontics. Am J Orthod 1981;80:
349-64.

APPENDIX hold on
Interactive computer program for Matlab 2014 plot(p,q);
software axis('equal');
xlabel('x')
% ELLIPSE - A script le to draw ellipse ylabel('y')
(x/a)^21(y/b)^251 and title('Ellipse (x/a)^21(y/b)^251 ')
% to calculate arch perimeter change by altering clear alpha x y ;
intermolar width and arch length. h 5 (AL-IW/2)^2/(AL1IW/2)^2;
alpha 5 linspace(-pi/2,pi/2,100); AP 5 0.5*pi*(AL1IW/2)*(11((3*h)/(101sqrt(4-3*h))));
IW 5 input('Enter intermolar width in mm: '); Nh 5(NAL-NIW/2)^2/(NAL 1 NIW/2)^2;
AL 5 input('Enter arch length in mm: '); NAP 5 0.5* pi*(NAL 1 NIW/2)*(11((3*Nh)/(101
CM 5 input('Enter the amount of intermolar change in sqrt(4-3*Nh))));
mm: '); % AP: Arch Perimeter, NAP: Arch perimeter after
NIW 5 IW 1 CM; expansion.
CI 5 input('Enter the amount of the incisor change in formatSpec 5 'An original arch perimeter is %6.2f
mm: '); mm yn';
NAL 5 AL 1 CI; fprintf(formatSpec, AP)
% CM: intermolar change, CI: incisor change, NIW: new formatSpec 5 'After the change, new arch perimeter is
intermolar width, NAL: new arch length. %6.2f mm yn';
x 5 AL*cos(alpha); fprintf(formatSpec, NAP)
y 5 (IW/2)*sin(alpha); formatSpec 5 'The net change of arch perimeter is
p 5 NAL*cos(alpha); %6.2f mm yn';
q 5 (NIW/2)*sin(alpha); fprintf(formatSpec, NAP-AP)
plot(x,y);

American Journal of Orthodontics and Dentofacial Orthopedics February 2015  Vol 147  Issue 2