THE VOLUME OF AN n-SIMPLEX
WITH MANY EQUAL EDGES
Stanley Rabinowitz
Westford, Massachusetts
It is well known that the volume of a regular n-simplex with
edge length s is r
sn n+1
.
n! 2n
But suppose one edge has length b and all the other edges have
length a. Is there a simple formula for the volume of the simplex
in that case? What if all the edges incident at a given vertex have
length b and all the other edges have length a?
It is these questions that motivated the investigation that led
to the following result:
Theorem. Let K be an n-simplex in E n . Suppose the vertices of
K are colored with r colors, c1 , c2 , . . ., cr (1 ≤ r ≤ n + 1). Let the
number of vertices colored ci be mi (1 ≤ mi ≤ n + 1). It is given
that if an edge has both its vertices the same color, ci , the length
of that edge is ai . If the two vertices of an edge have different
color, the edge has length s. Then the volume of K is
v
r u r
Y r
1 Y X mi
am
u
i −1 t
(−1)r+1 (m − 1)a 2 − m s2 .
n i i i i
n! 2 2 i=1 i=1 i=1
(mi − 1)a2i − mi s2
11
Proof. The volume, V , of an n-simplex in terms of the edge
lengths, {aij }, is determined by the formula
(1) (−1)n+1 2n (n!)2 V 2 = D
where D is given by the determinant
0 a212 a213 ··· a21n a21,n+1 1
a221 0 a223 ··· a22n a22,n+1 1
a231 a232 0 ··· a23n a23,n+1 1
.. .. .. .. .. .. .. .
. . . . . . .
a2n+1,1 a2n+1,2 a2n+1,3 · · · a2n+1,n 0 1
1 1 1 ··· 1 1 0
(See [1] for a proof.)
Now, let us assign the edge lengths as specified in the theorem,
except that to make the computations simpler, let us assume the
√ √
edge lengths are ai and s (instead of ai and s). A simple
transformation then will change the result we get into the form
required by the statement of the theorem.
We find that the resulting determinant consists of r square
blocks along the main diagonal and the last row and column being
the same as shown above. The ith block has the form
0 ai ai ai ··· ai ai
ai 0 ai ai ··· ai ai
ai ai 0 ai ··· ai ai
ai ai ai 0 ··· ai ai
. .. .. .. .. .. ..
. .
. . . . . .
a ai ai ai ··· 0 a
i i
ai ai ai ai · · · ai 0
12
and every other element in the determinant has value s. For ex-
ample, if n = 11, r = 3, a1 = a, m1 = 4, a2 = b, m2 = 5, a3 = c,
and m3 = 3, then the determinant is as follows:
0 a a a s s s s s s s s 1
a 0 a a s s s s s s s s 1
a a 0 a s s s s s s s s 1
a a a 0 s s s s s s s s 1
s s s s 0 b b b b s s s 1
s s s s b 0 b b b s s s 1
s s s s b b 0 b b s s s 1 .
s s s s b b b 0 b s s s 1
s s s s b b b b 0 s s s 1
s s s s s s s s s 0 c c 1
s s s s s s s s s c 0 c 1
s s s s s s s s s c c 0 1
1 1 1 1 1 1 1 1 1 1 1 1 0
We now proceed to evaluate this determinant by applying el-
ementary row and column operations. In each group of mi rows
(i = 1, . . . , r), we subtract every row (except the last row) from
the row above it. Then, in each group of mi columns, we subtract
each column (except the last column) from the column to its left.
We wind up with a matrix where each square block along the diag-
onal has been replaced by a matrix whose diagonal entries are all
−2ai , (except for the lower right entry with value 0), and whose
minor diagonals just below and above the main diagonal all have
value ai . Furthermore, all the s entries have disappeared with the
exception of those whose rows and columns are at the end of the
groups of mi . The 1’s in the last row and column have also turned
to 0’s except those occurring at the ends of groups of mi entries.
13
In our example, the resulting determinant is
−2a a 0
a −2a a 0
a −2a a 0
a 0 s s 1
−2b b 0
b −2b b 0
b −2b b 0 .
b −2b b 0
s b 0 s 1
−2c c 0
c −2c c 0
s s c 0 1
0 0 0 1 0 0 0 0 1 0 0 1 0
All missing elements in the display are 0’s.
In each square block, we can remove the ai ’s situated along the
two minor diagonals by adding in the appropriate multiple of the
preceding row or column, in succession, top to bottom and left to
right. In our example, we would multiply the first row by 1/2 and
add it to the second row, then multiply the first column by 1/2 and
add it to the second column. This leaves us with ( 21 − 2)a = − 23 a
in row 2 column 2. Thus, we multiply row 2 by 2/3 and add it to
row 3. Then we multiply column 2 by 2/3 and add it to column 3.
This leaves us with − 43 a in row 3 column 3, etc.
In general, the multipliers will be 1/2, 2/3, 3/4, . . . ,
(mi − 1) / mi . The final numbers along the main diagonal will
be −2ai /1, −3ai /2, −4ai /3, . . . , −mi a/(mi − 1), (mi − 1)a/mi .
14
In our example, we get
−2a
1 0
−3a
2 0
−4a
3 0
3a
4 s s 1
−2b
1 0
−3b
2 0
−4b
3 0 .
−5b
4 0
4b
s 5 s 1
−2c
1 0
−3c
2 0
2c
s s 3 0
0 0 0 1 0 0 0 0 1 0 0 0 0
Most of the entries on the major diagonal now have all 0’s in their
rows. We can thus expand the determinant by minors along these
rows and see that the value of the determinant is
r
Y
(−ai )mi −1 mi
i=1
times the following determinant
(m1 −1)a1
m1 s s ··· s 1
(m2 −1)a2
s m2 s ··· s 1
(m3 −1)a3
s s ··· s 1
m3 .
.. .. .. .. .. ..
. . . . . .
(mr −1)a4
s s s ··· mr 1
1 1 1 ··· 1 0
This determinant is simplified by subtracting s times the bottom
row from every other row. We are left with a determinant whose
15
last row and column are all 1’s (except for the 0 in the lower right
corner). The remaining elements all lie along the main diagonal,
(mi −1)ai
and are mi − s, i = 1, 2, . . . , r. In our example, this comes out
to
3a
4 −s 0 0 1
4b
0 −s 0 1
5
2c .
0 0 3 −s 1
1 1 1 0
Finally, this determinant is evaluated by getting rid of the 1’s in
the final row. To do that, multiply each of the first r rows by the
reciprocal of the diagonal element and subtract the result from the
last row. This changes the 1’s in the last row to 0’s and changes
the 0 to
r −1
X (mi − 1)ai
− −s .
i=1
mi
The determinant is now upper triangular and so its value is the
product of the diagonal elements. We have thus found that
r Xr
Y mi
(−ai )mi −1 (mi − 1)ai − mi s −
D= .
i=1 i=1
(mi − 1)ai − mi s
Pr
Comparing this with formula (1) and noting that i=1 mi = n+1,
we see that we can move the (−1)n+1 to the right hand side and
wind up with (−1)r+1 . Then, solving for V 2 and taking the square
root of both sides proves our theorem.
Letting r = 2 gives us two interesting corollaries.
Corollary 1. An n-simplex in E n (n ≥ 1) has one edge of length
b. Every other edge has length a. Then the volume of the simplex
16
is
ban−2 p
n 2na2 − (n − 1)b2 .
n!2 2
Corollary 2. An n-simplex in E n (n ≥ 1) has every edge incident
at a given vertex of length a. Every other edge has length b. Then
the volume of the simplex is
bn−1 p
n 2na2 − (n − 1)b2 .
n!2 2
Ref erences
1. D. M. Y. Sommerville, An Introduction to the Geometry of n
Dimensions. Dover Publications, Inc. New York, 1958.
17
