A Method of Estimating Plane Vulnerability Based On Damage of Survivors" by Abraham Wald (CRC)

AO-AO9 073 CENTER FOR NAVAL ANALYSES ALEXANDRIA VA OPERATIONS EV--ETC F/6 1/3
REPRINT OF 'A NETHOD OF ESTIMATING PLANE VULNERABILITY BASED -- ETCIU)

9 JUL go A WALD
UNCLASSIFIED CRC-432
smmhhhhmmml
EIIIIIEIIEIII
IIIIIIIIIIIImI
IIIIIIIIIII
ollllllllllll
Ot F
!9 0
A S T-
'"Afl 4 t Ali
:ON DAAE.O
ACAM
Abhp W
~~Ai-
1ww
ii itI , iW -R
NOW .il W, A ..,11

SECURITY CLASSIFICATION OF THIS PAGE (Mehn Data Entered)
REPOT DCUMNTATON
AGEREAD INSTRUCTIONS
REPOT DCUMNTATON
AGEBEFORE COMPLET04G FORM
IREPORT NUMBER 2.GOVT ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBER
CRC 432 &

4. TITLE (and Subtitle) S YEO EOT&PRO OEE
A REPRINT OF "A METHOD OF ESTIMATING PLANE

VULNERABILITY BASED ON DAMAGE OF SURVIVORS" Reprint - 1943
BY ABRAHAM WALD S. PERFORMING ORG. REPORT NUMBER
See Block IS
7. AUTHOR(@) 2. CONTRACT OR GRANT NUMMER(#)
Abraham ald
17 PER61RMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK
AREA II WORK UNIT NUMMERS
Statistical Research Group/National
Defense Research Commwittee,
Applied Mathematics Panel
I I. CONTROLLING OFFICE NAME AND ADDRESS 12. RFPORT DATE
Center for Naval Analyses ui'1980
2000 N. Beauregard Street 3 EAOF PAGES
Alexandria, Virginia 22311
14. MONITORING AGENCY NAME & AODRESS(il different from, Controlling Office) IS. SECURITY CLASS. (of this report)
Office of the Chief of Naval Operations COp03) ca iie

Washington, D.C. 20350 150. DECL ASSIFIC ATION/ DOWNGRADING
SCHEDULE
16. DISTRIBUTIqN STATEMENT (of this Report)
Approved for public release; distribution unlimited
17. DISTRIBUTION STATEMENT (of the abstract entered in &lack 20. it different froat Report)
IS. SUPPLEMENTARY NOTES

1) SRG Memo 85/AMP Memo 76. 1 5) SRG Memo 96/AMP Memo 76. 5
2) SRG Memo 87/AMP Memo 76.2 6) SRG Memo 03/AMP Memo 76.6
19. KEY WORDS (Continue on reverse side if neceeemry And ldentWf by block number)
aircraft, equations, kill probabilities, probability, vulnerability
20. ABSTRACT (Contineue on reverse side 11 necessary mid identify by block imober)
This research contribution consists of a series of eight memoranda

originally published by the Statistical Research Group at Columbia
University for the National Defense Research Committee in 1943 on methods
of estimating the vulnerability of various parts of an aircraft based on
damage to surviving planes. The methodology presented continues to be
valuable in defense analysis and, therefore, has been reprinted by the
Center for Naval Analyses in order to achieve wider diseemination.,r
DD FORM7 147 EDITION OF.I NOV 65...OBSOLETE

I AN) ' S., N 0102- LF- 014. 6601 SECURITY CLASSIFICATION OF1 THIS PEEf'(Ufin ZBate e
145-
LUj
0
z0
Z oMEMORANDUM FOR DISTRIBUTION LIST
0
X Subj: Center for Naval Analyses Research Contribution 432
CL
0
. Encl: (1) CRC 432, "A Reprint of 'A Method of Estimating
< Plane Vulnerability Based on Damage of Survivors'
< by Abraham Wald," July 1980
>N
> T I. Enclosure (1) is forwarded as a matter of possible
interest.
Z 4 2. This Research Contribution contains a series of
-memoranda written by Abraham Wald of the Statistical
Research Group at Columbia University during World War II.
Unfortunately, this work was never published externally,
although some copies of his original memoranda have been
available and his methodology has been employed in the
I I. analysis of data from both the Korean and Vietnam Wars.
-It is published by CNA not only as a matter of historical
interest but also because the methodology is still relevant.
3. Research Contributions are distributed for their

I II potential value in other studies and analyses. They do
not necessarily represent the opinion of the Department
0 of the Navy.
Z 0
(. PHIL E. DeP6Y
Director
Operations Evaluation Group
By--'' -
By ..... ....
. ..
Dis k A'
an affiite of the University of Rochester

Subj; Center for Naval Analyses Research Contribution 432
DISTRIBUTION LIST
Department of the Navy
SNDL
A2A CNR
B3 National Defense University (NDU)
B3 Armed Forces Staff College (AFSC)
SNDL Part II:
E3A NRL
FF38 USNA, Nimitz Library
FF44 NAVWARCOL
FKA1A COMNAVAIRSYSCOMHQ
FKR3C NAVAIRTESTCEN
FT73 NAVPGSCOL
OpNav: Op-09BH, Op-96, Op-05, Op-50W
Other:
Defense Technical Information Center (DTIC) (12 Copies)
Department of the Air Force (SAMI)
Air University
Air Proving Ground Center, Eglin AFB
Institute for Defense Analysis (IDA)
Mathematics Research Center
The Mitre Corporation
The Rand Corporation
University of Rochester (10)
Navy Tactical' Support Activity
LSRC-43 July
1/ 1980
1/
Aa REPRINT OF
---- AMETHODOF STIMATING
PLANE VULNERABILITY §ASED
_ ONQAMAGE OF 5URVIVORS''/
BY ABRA-HAM/AL D'.
Abraham Wald /1
Ll;
Operstions Evaluation Group
CENTER FOR NAVAL ANALYSESt

2000 North Beuregard Street, Alexandria, Virlginia 22311! 1
401
-o4li
Vr 104 A*r#10"
-- A:I t . . . 01 .f
$4%%~r-
l PAW-
*0*4..
tt b'St w
va4u Wtttj4U w~
n TV - Minimu OiC mci u Vine fl the. tAb4Lity

That, a- ns vifl beu e. bt -ay,iyf
OfHit. i~&ctat*4 1Mli*t $ome Ruttr
btber
Mnatrctsoaz on the:flobabiltltti q 1 .**q4- P 39
'Pat V Subdiviona Of *6i flM. fIto Seral Ewi-
port Vl ampling1. Etot#e,.t....u.4..es4.* 6

Part VI f&tk***Mfkutspttt*.sese 75
Part VIzM' VzZtbflIt$- of A Plane to Differnt typos'
7f7
4:li t
A 0r. *WA' o-,
A A
P ~ ~ ~ ~
*'A r" 4WfA7~AQ-
@4~qf
ir 0*ro -~r$Wufl ths m4,6taft
i~art Ito,
Aft ct
P.t4 njr~jp0y'' Th nasber.e
kpsaet gboe.
A,, flW4 tt o"~ tno bwe "*a. to accIect~aate cwt

"adto het&t*
tn%4tn from -nl Fart (uemarendwu) to:
Tba~iub.tpw. - 4v441t
ri wcfr41tg. booer, have be"
iu~ar
tfi wre aw.uir4 Aroflh tin fltina~l
Y'I
. :V;-
0A A
PART I
AN EQUATION SATISFIED BY THE PROBABILITfES THAT A

PLANE WILL BE DOWNED BY i HITS
INTRODUCTION
Denote by P. (i = 1,2,..., ad inf.) the probability that a plane
1
will be downed by i hits. Denote by pi the conditional prob-
ability that a plane will be downed by the i-th hit knowing that
the first i - 1 hits did not down the plane. Let Qi = 1 - Pi and
qi= - pi (i = 1,2,..., ad inf.). It is clear that
=
Qi qlq2"" "gi (1)
and
P1 = I - q 1 q 2 ... qi • (2)
Suppose that pi and Pi (i = 1,2,...) are unknown and our infor-
mation consists only of the following data concerning planes

participating in combat:
* The total number N of planes participating in combat.
* For any integer i (i = 0,1,2,...) the number A. of

1
planes that received exactly i hits but have not been
downed, i.e., have returned from combat.
A.
Denote the ratio R- by a i (i = 0,1,2,...) and let L be the
proportion of planes lost. Then we have
a 1(3)
i=0
IThis part of "A Method of Estimating Plane Vulnerability

Based on Damage of Survivors" was published as SRG memo 85 and
AMP memo 76.1.
.......
The purpose of this memorandum is to draw inferences concerning
the unknown probabilities pi and Pi on the basis of the known
quantities a0 , al, a 2 ,..., etc.
To simplify the discussion, we shall neglect sampling errors,

i.e., we shall assume that N is infinity. Furthermore, we shall
assume that
0 < Pi 1 (i = 1,2,..., ad inf.). (4)
From equation 4 it follows that
0 < Pi < 1 (i = 1,2,..., ad inf.). (5)
W;e shall assume that there exists a non-negative integer n such

that an > 0 but a i = 0 for i > n.
We shall also assume that there exists a positive inteqer m such

that the probability is zero that the number of hits received by
a plane is greater than or equal to m. Let m' be the smallest
inteqer with the property that the probability is zero that the
number of hits received by a plane is greater than or equal to
M'. Then the probability that the plane receives exactly m' - 1
hits is positive. We shall prove that m' = n + 1. Since an > 0,
it is clear that m' must be greater than n. To show that m'
cannot be greater than n + 1, let y be the proportion of planes
that received exactly m' - 1 hits. Then y > 0 and
y(l - pm,l ) = amo,. Since y > 0 and 1 - Pm'-I > 0, we have
am,_l > 0. Since a, = 0 for i > n, we see that m' - 1 < n,
i.e., m'( n + 1. Hence, m' = n + 1 must hold.
Denote by x i (i = 1,2,...) the ratio of the number of planes

downed by the i-th hit to the total number of planes
participating in combat. Since m' = n + 1, we obviously have
xi = 0 for i > n. It is clear that
n
xi. L 1 0 al an (6)
-2- el
is
CALCULATION OF x. IN TERMS OF a,0 a,,...,a,
np .I.,pn
Since the proportion of planes that received at least one hit is

equal to I - a0 we have
x Pi(I
1 a). (7)
hle proportion of planes that received at least two hits and the
firs.t hit did not down the plane is obviously equal to
II- aa -x. Hence, x) i23..r 8
In general, wci obtain

x.i p.( -a - a1 . - I- x2 .. i-I(9
Putting
c I - a0 - a1 .. - a. 1 (1u)
equation 9 can be written
*~ x+ p1 (xi + ... +x i-~=pc i=23..n.(11)
Substituting i - 1 for i, we obtain from equation 11
xi + P~iI(x I + ... + Xi 2 ) - P-c-

(i=34..n.(2
Dividing by pl_1 , we obtain
11 + (xI+ .. + x i- c
C- (i =3,4,...,n). (13)
1 i- i-
-3-
Adding x 1 -1 I x i-I to both sides of
equation 13, we obtain
+ q~i-1 14
1i-i i- - 11(4
From equations 11 and 14, we obtain
X.+ Pi x~~A~1 (15)
Hence,
d.i P.(c. - ci1 ) i (i= 3,4,...,n). (17)
and
t piqii (i = 3,4r ...,n). (18)

Spi-1
Then equation 16 can be written as
x d1 + t i-i (i =3,4,...,n). (19)
Denote pl(l - a.) by dl, -pa, by d2 and - by t; then we

-P2 2 P1 2
have
= d, and = ~ 1 + d2 (20)
-4-
Prom equations 19 and 20, we obtain
x- 1
X: j djtj+jtj+ 2 ti +d i (i = 2,3,...,n). (21)
EQUATION SATISFIED BY ql,...,qn
Ton derive an equation satisfied by ql...qn, we shall express

i xi in terms of the quantities t i and d i (i =
Substituting i for i - 1 in equation 14, we obtain

(22)
x =i
i (d.) + d
i L j -i p. i ' P i
Hence, in particular
n qn [n-i tn ) + (23)
Xn Z x = cn - n E (djtj+l'' d =]b= .
X
j=l
=1 [J=n
Pn
Since c L = an, and since t t ...V q .. qn- we
nnj+l n y q
obtain from equation 23
an - j ... + na. (24)
Dividing by q, ... qn and substituting -pjaj_ 1 for dj, we obtain
-5-
an a+ a ~ - n-I1 d
ql "'" qn ql "'" qn-I j=l pjql ... qj-1
an an- 1
q, ... q1 "'" q.n-
(25)
n- a_ 1
+ = 1q, ' qj-1
d1
Pl
n a |
= .. qj (I - a O ) =.0
j=l q.
or
1 -a . (26)
l q1 "' qj 0
If it is known a priori that q ... = qn' then our problem

is completely solved. The common value of ql,...,qn is the
root (between 0 and 1) of the equation
n a.
1=1 qJ 0
It is easy to see that there exists exactly one root between zero
and one. We can certainly assume that q1 - q2 I-... > qn" We
shall investigate the implications of these inequalities and
equation 26 later.
ALTERNATIVE DERIVATION OF EQUATION 26
Let bi be the hypothetical proportion of planes that would have

been hit exactly i times if dummy bullets would have been used.
Clearly b i _ ai. Denote bi - a i by yi (i = 0,1,2,...,n). Of
n
course, bO = a O , i.e., Yo = 0. We have J b i = 1. Clearly

j=O
-6-
.-. v
Yi - Pi b i Pi(ai + yl) (i = l,2, ...,n). (27)
Hence,
P. 1 - q, "'" qi ai
yi = O- a a. = -a.. (28)
n
Since Yi = L, we obtain from equation 28
y
1=1
n ai n
.. =q + a i =I - a . (29)
This equation is the same as equation 26. This is a simpler

derivation than the derivation o1 equation 26 given before.
However, equations 21 and 22 (on which the derivation of equation
26 was based) will be needed later for other purposes.
As mentioned before, equation 29 leads to a solution of our

problem ii it is known that q =n q In the next
memorandum (part II) we shall investigate the implications ot
equation 29 under the condition that ql > q 2 > ... > n"
NUMERICAL EXAMPLES
N is the number of planes participating in combat. A0 , A,, A 2 ,

...,An are the number returning with no hits, one nit, two hits,
...,n hits, respectively. Then
a. = A.
t-(i = O,1,2,...,n)
14N
i.e., a. is the proportion of planes returning with i hits. The

1
computations below were performed under the following two
assumptions:
-7-
loom
" The bombing mission is representative so that there is no
sampling error.
" The probability that a plane will be shot down does not
depend on the number of previous non-destructive hits.
Example I: Let N = 400

and AO = 320 then ao = .80
A1 = 32 a I = .08
A2 = 20 a 2 = .05
A3 = 4 a 3 = .01
A4 = 2 a 4 = .005
A5 = 2 a 5 = .005
We assume q = q2 q 5 = qi' where qi is the probability of

a plane surviving the i-th hit, knowing that the first i - 1 hits
did not down the plane.
Then equation 26,
n a.
j=l q1 0
reduces to
n a.
2
E - -= 1 -a 0
j=l q])
Substituting values of a.
1
.08 .05 .01 .005 .005

q q2 q3 q4 55
-8-
or
.200q 5 - .0}80 4
- .050q 3 - .010 2 - .005c - .005 = 0.
The Birge-Vieta method of finding roots described in Marchant

Method No. 225 is used to solve this equation (table 1). We find
q = qi = .851, pi = .149 where Pi is the probability of a plane
being downed by the i-th hit, knowing that the first i - I hits
did not down the plane.
x i equals the ratio of the number of planes downed by the i-th

hit to the total number of planes participating in combat. Using
i equation 9
X i l - ao -CAI - . . - al
= Vl -X x 2
2 - . - xi- )
1 1 0
(i = 2,3,...,n)
for n = 5, we obtain
x I = P)(I - a0 ) = .030
x = P2(I - a - - x = .013
xo 1 - 2 x 2 ) = .004
3 = P 4 (1 -
a
a - I 2 - a 3I1 -
-1 x - x2 - x3 ) = .002
x5 p 5 ( -a aO - a I - aa 2 -a -- a 4 x - x I -
-x x 2- - x - x )
5
4ap55(I .001
-
4 - 3 2 3 3 x).0
4
Example 2: Let aO = .3, a I = .2, a 2 = .1, a 3 = .1, a 4 = .05, and

a= .05. Then the following results are obtained: q = .7,
p = 1 - q = .13, x1 = .09, x 2 = .05, x 3 = .03, x 4 = .02, and
x = .01.
The value of q in the second example is nearly equal to the value

in the first example in spite ot the tact that the values a
(i = 0,1,...,5) differ considerably. The difference in the
values a i in these two examples is mainly due to the fact that
the probability that a plane will receive a hit is much smaller
in the first example than in the secona example. The probability
"that
- a plane will receive a hit has, of course, no relation to
the probability that a plane will be downed if it receives a
hit.
-9-
TALE 1
I. Assume q I -Y
.200 -. 080 -. 050 -. 010 -. 005 -.005

4.200 .120 4.070 +.060 *.0SS
.200 4.120 4.070 +.060 +.055 4.050 * A

4.200 +.320 4.390 .450 a
.200 4.320 +.390 *.4S0 +.SOS - A1
A
- 2 - - - 1 - .0990 - .9010
2. Assume q - .9010 - 2
.2000 -. 0800 -. 0500 -. 0100 -. 0050 -. 0050

+.1802 +.0903 +.0363 +.0237 *.0168
.2000 +.1002 +.0403 +.0263 +.0187 +.0118 - 8

4.1802 +.2526 +.2639 +.2615 0
.2000 +.2804 +.2929 +.2902 +.2802 - B1
ab
Y3 Y2 - .qOo - .042113 - .858807
3. Assume q - .858887 - 73
.200000 -. 060000 -. 050000 -. 010000 -. 005000 -. 005000

+.171777 +.070826 *.024758 4.012675 +.006592
.200000 *.091777 +.028826 4.014758 +.007675 +.001592 - C

+.171777 +.226363 +.219179 +.20092S 0
.200000 4.263554 +.255189 +.233937 +.208600 - C1
C
4 73 .859807 - .007632 - .851255
4. Assume q - .851255 - 74
.2000000 4.080000 -. 050000 -. 010000 -. 005000 -. 005000

+.170251 +.076827 +.022837 +.010928 +.005046
.2000000 +.090251 +.026637 4.012837 +.005928 +.000046 0D

+.170251 4.2217S4 +.211606 +.191058 0
.2000000 4.260502 +.248531 4.224443 4.196986 - DI
p
Y5 - Y4 " -" .051255 - .000234 - .051021
., .. . .~ .- ...
PART II
MAXIMUM VALUE OF THE PROBABILITY THAT A PLANE WILL BE DOWNED

BY A GIVEN NUMBER OF HITS'
The symbols defined and the results obtained in part I will be

used here without further explanation. The purpose of this
i
memorandum is to derive the least upper bound of Xi L xj and
j=l
that of P (i = l,...,n) under the restriction that

i
q,--> q2 > '''> qn"
First, we shall show that X i is a strictly increasing function

of pj for j < i . Let us replace pj by pj + A (A > 0) and let
us study the effect of this change on xl,...,x i . Denote the
changes in x 1 ,...,x i by Al,... ,Ai, respectively. Clearly,

A1 = ... = Aj_ 1 = 0. It follows easily from equation 9 that
Aj > 0 and
Aj+l = -Pj+l Aj
Hence,
Aj + Aj+I = (1- Pj+I) Aj > U.
Similarly, we obtain from equation 9
-Pj+2(Aj + Aj+) =-Pj+2(l- pj+ ) Aj •

Aj+2
Hence,
A + Aj+ 1 + Aj+ 2 = (1- Pj+2) (I - Pj+ 1 ) Aj > 0.
In general
)
Aj + Aj+ 1 + ... + Aj+k = (1 - Pj+l) ... (I - Pj+k Aj > 0
(k = ,...,i-j)
Hence, we have proved that X i is a strictly increasing function

of P. (j = ...,i).

AMP memo 76.2.
-Il-
On the basis of the inequalities pi> pi-l' we shall derive the
least upper bound of X i . For the purpose of this derivation we
shall admit 0 and I as possible values of p (i = l,...,n), thus
i
making the domain of all possible points (Pl,"'pn) to be a
closed and bounded subset of the n-dimensional Cartesian space.
Since X i is a continuous function of the probabilities plV p 2 '
etc. (X is a polynomial in pl,...,pi), the maximum of X i exists
and coincides, ot course, with the least upper bound. Hence,
our problem is to determine the maximum of X.
First, we show that trio value of X.

I is below the maximum if
Pn > pi. Assume that pn > pi and let k be the smallest positive
integer for which pk> pi" Obviously k > i. Let p! = I; (I + E)
ior j 1.I....k-1, and p' = p (I - TI) for j = k,k+l, n,
o j
where £ > 0 and n is a function ri(£ ) of c determined so that
n
. x' = L (x' is the proportion of planes that would have been
I :x I
brought down with the j-th hit if p '•'''Pn were the true
n
probabilities). Since Xr (r = l,...,n) is a strictly monotonic
iunction ot pl,"'Pr' it is clear that for sufficiently siuall
such a function n( c ) exists. It is also clear that tor suffi-
ciently small c the condition p' < PK- . (
< '
Since p! > p. (j = l,...,i), we see that X' > X. (X i does not
depend on Pr* for r > i). Hence, we have proved that if
PIP'.'Pn is a point at which X i becomes a maximum, we must have

=
Pi= pi+l "- = Pn"
Jow we shall show that if X i is a maximum then p1 = . " =

For this purpose assume that pi > p1 and we shall derive a con-
tradiction. Let j be the greatest integer for which pj= pl.
Since pi> pl, we must have 3 < i. Let p' = pr(l + c) for
r = l,...,j and Pr = Pr(l - n ) for r = j+l,...,i, where
i i
xk
£ > 0 and n is determined so that X; = Then [or the
probabilities pi,...,pi, Pi+l....n the proportion ot lost
-12-
°
-*
planes is not changed, Now let pr =P
i.e., it is equal to L.
r I
for r > i. Then the proportion L' of lost planes corresponding
to P ,..,pA is less than L. Hence, there exists a positive
A so that the proportion L" of lost planes corresponding to the
probabilities pr = p. (I + A) is equal to L. But, since p" > p
i i i
(r = l,...,i)we must havejL x! > L x = Lx Hence, we
j=1 I j=l I j=l j
arrived at a contradiction and our statement that p1 = P 2 =

Pi is proved. Thus, we see that the maximum of X i is reached
when p1 - p 2 = ... = pn
LEAST UPPER BOUND OF Pi
Now we shall calculate the least upper bound of Pi. Admitting

the values 0 and 1 for pj, the maximum of Pi exists and is equal
to the least upper bound of Pi. Since P i 1- ...
maximizing Pi is the same as minimizing q 1 ... qi. We know that

ql,..,qn are subject to the restriction
j= - I - a . (30)
Z q . qj 0
Let qo,...,qn be a set of values of q,...,qn (satisfying

equation 30) for which q, "'" qj becomes a minimum. First, we
0 0 0 0 0
show that q. - i+l qn Suppose that q < qi
Consider the set of probabilities qrl qr for r < 1 and q'

r I
for r > i. Then
qI ... q j 0
-13-
Hence, there exists a positive factor A < I so that
n a.
=1 a 0
j=l 1
where 1= Aq. (i 1,...,n).

1 Then
q~q ... < qoq0 ... q0

0 .
in contradiction to our assumption that q " q? is a minimum.

o 0
Hence, we have proved that q= = q
Now we show that there exists at most one value j such that
1> qo > qo Suppose there are two integers j and k such that
) 1
>
I> qo> q. qO Let j' be the smallest integer for which
= qJ0 and let k' be the largest integer for which q = q°
q
Let qj, = (1 + £) qj,,
0 1 q 0 , (E > 0), and q =
j k' IT+ k r =r
for r$ j', k'. Then
n a
q . 0. "" qi and -< I - a0
r=l q1 ... q
q[
Hence, there exists a positive factor A < I such that
n ar a
r=±1 r -
<
where q= b ut q* ... qt < ql "'" i= ... qo, which
contradicts the assumption that q0 ... gO is a minimum. This

proves our statement.
-14-
II
"" ....... ,.............................
-"'" .......
'" ....
......
°"' .... " "-'"-' -i.................. '....... ' . - : ... ,"
It follows from our results that the minimum of q is the root of
the equation
r=.~r=i a0 (32)
Now we shall calculate the minimum of q~q 2. First, we know that

q q2 (i > 2) if q q2 be a minimum. Hence, we have to minimize
i 2 2 1
qlq 2 under the restriction
_/ a a a
q~d 1
+1 +
2
~n . qn-l
1 1- a 0 . (33)
2 2 q2 )/
Using the Lagranhge multiplier method we obtain the equations
__( a2 a an
q2 j (a + 2 3 +2 + n 0 (34)
(Lagrange multiplier = X)
Al(a 2 + 2a3+ (n - )an

q 1* 2+ 2a3 n n 0~ (35)
2 q22
Because of equation 33, we can write equation 34 as follows:
(1 -a)=0; A l 2
Substituting for A in equation 35, we obtain
- I a 2a 3 +j4+ +(n - l)an

1 2 + 3 * =-l0
0o q2 q2 q2 /(36)
-1.5-
. . .. . . .. .'
or
q( 1- a2 2a3
2 + + (n-n- 1)a n (37)
0 q2 q 2q 2 -
'
On the other hand, from equation 33 we obtain
q1 =
o
a2
1 + aq2 +
q2
a___n__
a, + ... +
q2
)- (38)
Equating the right-hand sides of equations 37 and 38, we obtain
a3 2a 4 3a 5 (n- 2)an
2 + 4 n-i - a1 = 0. (39)
q2 q2 q2 q2
It is clear that equation 39 has exactly one positive root. The

root is less than or equal to 1 if and only if
a 3 + 2a4 + 3a 5 + ... + (n - 2)a n < al (40)
Equations 38 and 39 have exactly one positive root in ql and q 2 "

We shall show that if the roots satisfy the inequalities 1 > q > q2'
then for these roots q 1 q 2 becomes a minimum. We can assume
that 2 < n, since the derivation of the minimum value of q, ... qn
will be given later in this memorandum. It is clear that for any
a1
value q, > T- equation 38 has exactly one positive root in
0
q2 " Denote this root by O(q1 ). Hence, O(ql) is defined for
a1
all values q, > I - a It is easy to see that
-16-
t.
'.. ,......................................................................................."......
. ..... ..
, .
•li ra (q i + o"
Hence (assuming
a1 a I > 0)
lir (q) =+
°
q,
where )
*(q = ql ONi )
It is clear that lira O(q )

= 0. Since a n > 0, it follows from
ql
equation 38 that ql [ n -
O (ql)] has a positive lower bound when
qlc. But then, since n > 2, li0) q, 0(ql ) +
c" From
ql -PC
the relations lim (ql ) = lim )
O(ql = + 0 it follows
q,al ql,
that the absolute minimum value of
*(ql ) is reached for some
positive value q Since equations 38 and 39 have exactly
positive root in q and q2' one
the absolute minimum
i(q value of )
must be reached for this root. This proves our statement that
of equations 38 and 39 satisfy the inequalities
if
q 1 roots
1 the > q2' then for these roots qlq 2 becomes a minimum con-
sistent with our restrictions on q, and q 2 " If 1 > q, > q is

2
not satisfied by the roots of equations 38 and 39, then qI is
equal either to I or to q 2 and the minimum value of q 1 q 2 is
either *(I) or q 2 , where q is the root of the equation
-17-
. .-
.....
.
.
.~
n a
r o
now we shall determine the minimum of ql' qi (2 < i < n).

First, we determine the minimum Mil of ql qi under the re-
i-I
striction that q = qi. Thus, we have to minimize qlq 2 under
the restriction that
S+ 2 3 + +
n I - a (40a)
+ + 2 n-I 0
ql qlq 2 qlq 2 qlq 2
Using the Lagrange multiplier method, we obtain
q __ a
_ + ** + n
aa/0o ~ i-1l(-
-qq qlq n-i 2 q (41)
2
and
(i - 1)qlq i-2
2 -
x
. (9
a2
2 +
2a3
+ +
(n - 1)an
+'" 0
1 q2 (41a)
qlq2-
Substituting 1 - a for X (the value of X obtained from
0
equation 41), we obtain
(i - 1 1
- aO a2+23+(
+ 2- - + + n-I )an 0
-ql
0 q2 q2 q / (42)
From equation 40a

a2 an
(i 1)q a + + + n= 0 . (43)
Iqa
-
2\ qf /
-18-
°,
L-------------. . .. "'-. . '5 '5"" '.:, - •,
From equations 42 and 43, we obtain -
(i - 2)a 2 (i - 3)a 3 (i -n)an

IaI+-+ 2 + + n-1I0
1 q2l~ q2 q2(44)
From Descartes' sign rule it follows that equation 44 has exactly

one positive root.
0
Let q,= q, and q2= q 2 be the roots of the equations 43 and
o 0 i-I
44. If I > q - qo ,then Mi. q (q ) .If I > qo > q'o does
not hold, then M lis either ()'or (qw" i-1 where q' is the I
root of the equation
3- =1-a (45)
j(qS)J
and q" is the root of the equation
a 2 +a3 2+ ,+ ann1a0(6
(q11 2 (q")n = - .(6
qf
Let M.i (r = 2,...,i-1) be the minimum of q1L ... qiunder the

restriction that q-l q~ I n= q1. Then M. can
be calculated in the same way as Mi,; we have merely to make the
substitutions
n n- r
n +I
=o 0 +a I + .+ ar-l
a. = a.jrl1,,,
qj qj+r-1 (
i ii- r +
and we have to calculate the minimum of q, . qi . Thus, we

have to solve the equations corresponding to equations 43 and 44,
i.e., the equations
(i - )q* + 2 + + n 0
aoq2 (q2 ) (q 2 )
(43*)
and
* * (i2)a 2 ( -3)a ( n)an_
n
(i - )a 1 + 2* 3 +2 -n0.
q(ql (q 2 ) 1
2 (44*)
2 2
Le tqt v1 and q2= v 2 be the positive roots ot the equations
43* and 44*. If 1 > v1 > 2 then Mi vv ' If 1 >v >Lv2
does not hold, then M ir is equal to either (Vl) 1

or f
i -
(v") ,where v' is the positive root of the equation
nE a, = 1 - a *(45*)
3=1 (v')l 0
and v" is the positive root of the equation
a2 a3 a **
a1+- + -~ + 0.. - I- a .(46*)
1 v" (v") 2(v") n-
The minimum Kiof q, ... qi (i = 2,3,...,n-1) is equal to the

smallest of the i - I values M ,...,"M.l-
flow we shall determine the minimum of q. ... q We show that

the minimum is reached when q, . qn~l 1. Suppose that
this is not true and we shall derive a contradiction. Let j be
the smallest integer for which q < I (j < n). Lot . (1+ C)q;
(c~ ~ = nO,
(c 0) q 1 and q for all r ,'n.
Then 4... 4n q 1 *09 q nand
-20-
n a
r=l q, ... qr
Hence, there exists a positive X < I such that

n ar a,
ql 0.. q
where
qr =
r
* *
But then q ... q< q, ... = q ' qn in contradiction to

the assumption that q, ... q is a minimum. Hence, we must have
q - = qn-l = 1. Then, from equation 26 it follows that the
minimum value of q, ... qn is given by
an
-a o -a- an-i
If i > I but < n, the computation of the minimum value of q, ... qi

is involved, since a large number of algebraic equations have to
be solved. In the next part we shall discuss some approximation
methods by means of which the amount of computational work can be
considerably reduced.
-21-
PART III
APPROXIMATE DETERMINATION OF THE MAXIMUM VALUE OF THE PROBABILITY

THAT A PLANE WILL BE DOWNED BY A GIVEN NUMBER OF HITSI
The symbols defined in parts I and II will be used here without

further explanations. We have seen in part II that the exact
determination of the maximum value of P 1 (i < n) involves a con-
siderable amount of computational work, since a large number of
algebraic equations have to be solved. The purpose of this
memorandum is to derive some approximations to the maximum of P 1
which can be computed much more easily than the exact values.
Let us denote the maximum of P. by pO and let Qi = I - PO
Thus, Q0 is the minimum value of Qi. Before we derive approxi-

1 1*
mate values of IP (or Qi) we shall discuss some simplifications

1 1
that can be made in calculating the exact value P0 (or Q)
1 1
assuming 1 < i < n. We have seen in part II that Q is equal to

the smallest of the i - I values MwMi We shall
make some simplifications in calculating Mir (r =
For this purpose consider the equation

ar ar an
r + r+l + ... + n 1-a -a -. ar (47)
u uv uv n-r 0 I-l
a
It is clear that for any value u >1 r .. aoar_
47 has exactly one positive root in v. Denote this root by *r(u)

a
is defined for all values u > r
Thus, r (u)
r I - a0 - ar-l
In all that follows we shall assume that a i > 0 (i =
We shall prove that
lim ([,u]i-r' =
U. r~u G (48)

AMP memo 76.3.
-22-
Vj ,'. . : , _ . _ . . L . , _ _ ... - - , .i +
and
ui r(u) ]ir +0G (49)
u
It follows easily from equation 47 that if u-* -a -r
o .. "- ar-l
then #r(u) - +w. Since i > r, we see that equation 48 must
hold. It follows easily from equation 47 that lim *r(u) = 0.
U=+ OD
r(u n- r
Vie also see from equation 47 that if u-*w, the product u
must have a positive lower bound. Equation 49 follows from this
and the fact that lim *r(u) = 0.
U-0
We have seen in part II that equations 43* and 44* have exactly
one positive root in the unknowns, q and q 2. Let the root in
* 0 * o
q be u0. Then the root in q 2 is equal to # (u0).
I ir r ir
From equations 48 and 49 it follows that u [*r (u) Iris
a
strictly decreasing in the interval 1-a - aI u
U< U
- - a. ~ i
0
and is strictly increasing in the interval u. < u < +o.
Denote by u'
r the positive root of the equation
ar dr+l an
+ +
+ = ... n-r+l I - ao - ... - a rI. (50)
u u u
It is clear that ur < I and *r (u') = u' The value Mirs

equal to the smallest of the three values
O
u ri#~r
[l)]iur, [l)i-r,
[r and u?r rir r)ir "
[r(u
A simplification in the calculation of M. can be achieved by

the fact that in some areas Mir can be determined without
calculating the value u.

0 We consider three cases.
Case A: u r ]ir < #
-23-
Amu
-&
L7.
In this case,
ir
= u'
r
IrlU)i-r if U 0(u) i > 0 for u = u'
-- r
and
Mir =U r(Ur) if u (ujir < 0 for u u
Case B: Ur r(ulir > Or, 1 )]i-r
In this case,
Mir = r(1) if U u [r(U < 0 for u
and
M. u r(uir)J if u rUj > 0 for u = 1.
Case C: . (u,1li-r - 0i) -
In this case,
M
= 0
M.
r ur . jjYlu
o~i-r
0
ir I
We can easily calculate the value of --- u ru for u =u

du Lr r
and u = 1. In fact, we have
-i [r(u) [Or(u) + (i - r)u r(u)) d (51)
d~ru) dv
and - u- =- can be obtained from equation 47 as follows.
-24-
Iar
Denote -+
ar. 1
- + .. + -jj
an
by G(u,v). Then
-
rr
d r (u) dv__~ G(u v)

dudu L'~,v)
(52)
1(ar + lan
+- + **+ n-r
u u uV /
r+1 r+2 (n
- + -3 + . . n-r+l
vv
+ V3 n-r+l/
on the biasis of equations 51 and 52, we can easily obtain the
# (1)
value of d
diu [,(u)]i-r
rin for u =u,
r and
and u = 1 if u, r
have beens cal'culated. if u = u', then *r(u) = V - US'; if

r r
u - 1, then v = #r l).
Since #r (1) is equal to the root of the equation in v
an ~ a
4.-i_ .+ + -r 1 a -al - . ar-
v vn-o 1
it follows from equation 50 that
r(1)inU41 (53)
-25-
Thus, for carrying out the investigations of cases A, B, and C
for r = l,...,i-1, we merely have to calculate ui ...,u'.
If we want to calculate Q for all values i < n, then it seems

1
best to compute first the n quantities u ,...,u
Since u(
r Yr and we can say that Mir is
the smallest of rr)
the three values u'~,
(u1 .)ir+1, (ur+)ir, and Uir [r(uir)]r
Since Qi0 is equal to the minimum of the i - I values,

Mil'''Mi i-l' we see that
O< ti ' (54)
where
ti = Min
1.2.,
(uP)i, (u )i-.. (ui_
(U! 1 2 ,'(
V U! 5
(55)
2o
If n is large, it can be expected that Q will be nearly ecual to

1
0
ti . Thus, t i can be used as an approximation to Qi " In order
to see how good this approximation is, we shall derive a lower
bound z. for Q0 " If the difference t. - z. is small, we are
certain to have a satisfactory approximation to Qi if t i -z
is large, then t. still may be a good approximation to Qo0, since

1
it may be that z. is considerably below Qi "
0
To obtain a lower bound z i of Qi' denote by y" (j = Ol,.,i-l)
the proportion of planes (number of planes divided by the total
number of planes participating in combat) that would be downed
out of the returning planes with j hits if they were subject to
i - j additional hits. Then
+
Pi *Yo + y + Yi- + x. + x 2 + "'" + xi " (56)
-26-
It is clear that ajPi> y. (j = O,1,...,i-l) and consequently
(ao + aI + .. * + ai-l) P > Y. + Y, + + Yi-I
Hence,
Y + yI + * ++ < Pi (57)
a0 + aI + ai-l
i+
Equation 56 can be written
(ao + ... + a Y + yl + "" + Yi-

i =
P.
i-l) a0 + + ai_ 1
(58)
( - a0 - - a. _ x a+
1- 0. + x.
* . o - -9i_
i.
0i a0 a-
y. + .". + Yi-l
Hence, P. is a weighted average of YO + + ai-l and
. 0 a + + ai-
1. 1 Then, from equation 57 it follows that

1 -a - ... -ai_ 1
Px + . . + i5
i 1- - a I - ... - a (59
Since yj > 0, we obtain from equations 56 and 59
x + ... + x(
1 + 9"0 + xi < pi < I-a 1 i (60)
Hence,
1l - + .. + - ai_ Q <
o 1 - (x + + x1 ) (61)
-27-
In part II we have calculated the maximum value of x I + ... + x.-
Denote this maximum value by A i . Then a lower bound of Q i is

given by
A.
zi 1 - ao - ... a_ <- (62)
NUMERICAL EXAMPLE
The same notation will be used as in the numerical examples for

part I. qi is the probability of a plane surviving the i-th hit,
knowing that the first i - 1 hits did not down the plane. Then
the probability that a plane will survive i hits is given by
Qi = qlq 2 - qi
In part I it was assumed that
q, = q 2 = ... = qi = q0 (say),
which is equivalent to the assumption that the probability that a

plane will be shot down does not depend on the number of previous
non-destructive hits. Under this assumption
Qi = qi
io
~ 0
The example below is based on the assumption that
qlI q 2 -> ".' > qn'
i.e., the probability of surviving the i + 1 hit is less than or

equal to the probability of surviving the i-th hit. In this
case, it is not possible to find an explicit formula for Q,, but
a lower bound can be obtained. That is, a value of Qi can be
found such that the actual value of Q.
1 must lie above it. The
greatest lower bound is denoted by Qi " Hence, we have
O
Qi < o i
If
0
Pi - I -0 0 ,
-28-
bad""
0
PCo is the least upper bound of P.; that is, the probability of
being downed by i bullets cannot be greater than Pie

0 1
Since the computation of the exact value of Q0 is relatively

complex, an approximate formula has been developed. This
approximation is called ti and ti > QO. Another approximation
0
(zi ) is available such that zi 0 However, z. is not as

accurate as t.. Whenever the full computation is to be omitted,
I
it is recommended that t. be used.
1
The observed data of example 1, part I, will be used. Thus,
aO = .80, a, = .08, a 2 = .05, a 3 = .01, a 4 .005, a5 = .005
The calculations are in three sections:
" 0
The calculation of ti > Qi'
" The calculation of z Q
0
" The exact value of QO.
1. Calculation of ti (t.> QO)
(1) Calculate u',

r the positive root of equation 50:
rI
ar ar+l an
u +
7u 2 +
00 + n-
r+l = 1- a - ... -ar_ .
For r - 1, we obtain
a I + a22+ a33+ a1
4
-T+ a 55 aO0
u U u U u
-29-
......
..
which reduces to
.20u 5 - 08U 4 - 05u 3 .1

Ol 2
05OOu- .005 =0
u' = .851
For r = 2,
a2 a3 a4 a5
U U U U
which reduces to
4 3 2
.12u - 05u - Olu -005u =0
u2 = .722.
For r =3,
a3 a a5
-+- + -- la a a
U U2 U302
which reduces to
.07u3 - .0h
l 005u -. 005 = 0
u;= .531
For r = 4,
a4 a5
-+ 2~=1 a0 - a1 -a 2 - a3 ,
U U
which reduces to
.06u 2 .005u - .005 =0
u 4 =.333
(2) t 1 .... ,t 5 are given by equation 54:
-30-
t Mmi 1
(up), (up'1 1 ... 2,
(u!) (up]
We have
ul .851, u! .722, u. .531, u- .333
Hence,
t1=Min [(U!] =U
= .851
t2= Min [(up) 2

1 (ul)]
*= Min [.724, .722]
= .722
*t = Min [(up 3 (u2) 2 (uP1
33 2
= Min [.616, .521, .531]
= .521
4
t
4
=Min
I (u (u2, (u3. 2 , (u4.)]
= Min [.524, .376, .282, .333]-
= .282
t5is not calculated since the exact value of 5can be

easily obtained.
2. Calculation of z. (z. Q0
L1 i i i
The following values must be obtained:
q0,the root of equation 26A
a1 a2 a3 a4 a5
q + 77+ -+ -+ -9. I -a 0
q q q q q
This has already been obtained as ul Thus qo .851. The

values of x1 POS,x 5 have been calculated in part 1:
-31-
.-~*
..
. ' i..
x, .030, x 2 = .013, x = 04, x .002, x5 = .001.
AA1 : x1 + x 2 + *.. + X..1
A, = x, = .030
x
A = +x = 043
2 1 2
A3 = x + x 2 + x3 =.047
A 4 = X1 + x2 + x 3 + x 4 .049
A = x I + x2 + x3 + x + x5 .050.
5
From equation 62 the lower bounds z i are calculated:
A. i-:
zi = 11 - a - - ai 1 Qi
Then
A1 _ .030
z = 1 1-a 1 .0= .850
o '
a a.0
0
A3
0
z33 1 -1 - a=
aO - a -a a 2 1 =4.329
I .07
=1 A4 .049 = . 183
1 - '04
4 1 - ao - a1 - a 2 -a 3 .06
z is not calculated since Q5 can be obtained directly.
3. The Exact Value of 00
We have calculated t. and z i such that
-32-
1- 0<
i1-<t t 1 i-12..5
0
The exact value of Q. is obtained as follows:
M. - Min ( ,)i-r+l i i-r F o i-rl
}
ir Ur'(r+l) 'ir [fr(Uir
0 0
where u. and 0r(u.r) will be defined below.
ir r ir
Q = Min [Mil,...,M i , i -l ]
or combining theseo1equations with the definition of t. we obtain
O = Min It .851
22 = Min it22'
, u20
211 [ (u21 2)]h o
3 = Mn ft 3 , u3 1 [ *l(U3 1 )] 2 u32 [ * 2 (u32 )]}
oo o 2 o 0
4= M tV
4 , u4 1 * 1 (U41) , u42 [ 0 2 (u4 2)J , u 3 0 3 (u43)]}
*
If Ui0 > 1, F (u0)] > I, or r < (Ui), then

LrrlrJ Uir < r~ir
r (uU) is neglected in the equations above.

Qr0 =o .i a5 =.005
a5.0 = .091.
5 f1 -a a 1 - a2 - a 3 -a .055
In the equation of the additional quantities we have to

compute are
0 0
u 21
31 A(U2.
032 2U320
u41 .l(U41)
42 2 42
u42 (u42
u43 3u43)
-33-
The following equations have exactly one positive root in q* q*.
1 2.
O in qo2hiss
root inq
The root in q* is u ir;; the roo (ir)0
(u
a*2 a*
3 n____
,a* + + + * + -In (1 a *)q*
1 q2(q) 2(q*)nl
1 * ~ 2 o2
where q* satisfies
(i* - 2)a* (i* - 3)a* (i* -n*)an

(i* - l)a* + 2 + q*2 3 +.. 2
(q*) +
-In
(q*) n* 0,
where
n* =n- r+ 1
a* =a + a + .. + a
o o r-i
a j+r-I
S=i - r + 1
The details of the computation are given in tables Z and 3.
TABLE 2
u i r n* 1* a* a* a* a*
_ir o 1 2 3 4 5
u 2 1 5 2 .80 .08 .05 .01 .005 .005
u31•
o 3 1 5 3 .80 .08 ...05 .01 .005 .005
u 3 2 4 2 .88 .05 .01 .005 .005
U41•
o 4 1 5 4 .80 .08 •
.05 .01 .005 .005
u42
o 4 2 4 3 .88 .05 .01 .005 .005
u4 3 4 3 3 2 .93 .01 .005 .005
where
a0 = .80, a 1 = .08, a 2 = .05, a 3 = .01, a 4 = .005, a 5 = .005
-34-
3v
A
I-
a 0 0
on
o
*.4
400
10c
91
a. 0
8 a
s. +4 am4
aI I 9
a,4- air .ro aq am;
I ~ a oftI OS
o on0
,_I.o
-o ,V *,. o.o
a- + -
* 4
u ~ 9 9- 0.2 '
* - -* 4
-35-
O N
r I
ft
-"
, 4
44 U
++
|6 OOm
Off"
S0
00
-- 36
S-"a
*" ml
a 1' . -
Substituting the values from table 3 in equation A and neglecting
several terms as explained in table 3, we have
0.
Ql = .851
0 = Min 1.722, .7211 = .721
Q = Min 1.521, .5171 = .517

0 = .282
0 = .091
The results obtained are shown in table 4.
TABLE 4
i ZQ t. q
1 .851 .851 .851 .851

2 .642 .721 .722 .724
3 .329 .517 .521 .616

4 .183 .282 .282 .524
.5 -- .091 -- .446
Thus, with the observed data, this example, if all the

information available about the qi's is that
>
q, I q 2 ... > q5 '
all we can say about the Qi is that
01 > .85, Q2 > .72, Q 3 > .52, 04 > .28, 05 - .09
Note that
Z Q' = tl = qo °
-37-
This is always true.
It is interesting to compare Qi with the values of Q. obtained

under the assumption that all the qi's are equal and have the
value qo. Under this assumption,
i
Qi = q (i = 1,2,...,5).
o o o 3
In table 4, 01 = qo and Q2 is very close to q 02 Q3 and q0
differ by approximately .1 and the agreement between Qio and
i
qo gets progressively worse. It will usually be true that
0
and Q are approximately equal for small values of ii but will
differ widely as i increases.
-38-
-e
PART IV
MINIMUM AND MAXIMUM VALUE OF THE PROBABILITY THAT A PLANE

WILL BE DOWNED BY A GIVEN NUMBER OF HITS CALCULATED UNDER
SOME FURTHER RESTRICTIONS ON THE
1
PROBABILITIES q,'''''q
In parts I, II, and III we merely assumed that q, > q q

In many cases we may have some further a priori knowledge
concerning the values qlo,,qn. We shall consider
<
here the case when it is known a priori that lqj< qj+l 2 qj
< <
(j = l,...,n-l), where XI and (I )2 12 are known
1)
positive constants.
We shall also assume that

n a .L
< - a (63)
j ft L o
l
Since a I + a2 + ... + a n < 1 - a0 , the inequality in equation

63 is certainly fulfilled if A is sufficiently near 1. It
follows immediately from equations 63 and 26 that q 1 < 1.
CALCULATION OF THE MINIMUM VALUE OF Q. = I - Pi (i < n)
Let 0 be the values of qlo,.r,q which Qi becomes

a minimum. We shall prove the following.
Lemma 1: The relations
q0 ql= xq0
A2 qj (j
j- i,...,n-l) (64)
(4
must hold.
Proof: Suppose that the relation in equation 64 does not hold

for at least one value j > i and we shall derive a contradiction.
1
This part of "A Method of Estimating Plane Vulnerability
AMP memo 76.4.
-39-
LetLtqr
r r - l,...,i and q+
qofrfor r'n - A q!3 ffor j = i,...,n-1.
Then we have
n a.
q.. q!= qL..and.
o < I.- a .(65)[
Hence, there exists a positive value A < I such that
n a.
z= 1 =j
where q! = Aq! (j = l,...,n). But then

I I
q u .. . q ! < q j .. . q l q l " ' " q 0i
0 0
in contradiction to our assumption that q.' q? is a minimum.
Hence, Lemma 1 is proved.
Lemma 2: If j is the smallest integer such that qk~1 = A2qk for
all k > j, then q = Alqo- for r = 2,3,...,j-l.
Proof: Assume that Lemma 2 does not hold and we shall derive a
contradiction. Let u be the smallest integer greater than one
0 o
such that qu > Xlqu_ 1 it follows from the definition of the
integer u that if u > 2, then q-= Alqu
0 2 From assumption 63
it follows that q, < 1. Hence, if we replace qu-I
qu_- (1 + c)q°_l (c > 0), then for sufficiently small c the

inequalities AXqr < qr+l< X2 qr (r = l,...,n-l) will not be
disturbed. Let v be the smallest integer greater than or equal
0 0
to u such that < A2 Since by assumption j is the
smallest integer such that qk+l = X2 qk for all k > j, we must
have qj < A qj_, . Hence, v < j-l. It is clear that replacing

0
qby qv 1 + we shall not disturb the inequalities
Alqr < q < A2 qr (r - 1,...,n-1). Hence, if
-40-
0
u-I (1q+u-i'qv q and q' o

=u- c ' qr "
- A2 q I n-1) is ful-
for r P(u, / v, then lq _jqk+l (k
filled. Furthermore, we have
=qoj..q
. 0 and aa <
S" ii j=l qj "'" 0
Hence, there exists a positive A < 1 such that
n a.
j -- ql ... q" o
and q" = Aq! (j l,...,n). But then

I I <
qj ...q! ql ...q! =-q"" q0
in contradiction to the assumption that q0 " is a minimum.

Let Eir (r = l,...,i-l) be the minimum value of Qi under the
restriction that qj+l = X2 qj for j = r+l,...,n-I and q,+, = Xlqj

for 3 = l,...,r-l. From Lemma I and 2 it follows that the mini-
mum of Qi is equal to the smallest of the i - 1 values Eil,...,Ei'i_ I .
The computation of the exact value of Eir can be carried out
in a way similar to the computation of Mir described in part
II. Since these computations are involved if n is large, we shall
discuss here an approximation method.
Let E (r - l,...,i-l) be the value of Q. if q+l A q for
jIl 2qj
j - r+l,...,n-I and qj+l Alqj for j - l,...,r.
A Furthermore,
let 0 be the value of 0. if q X q (j Then,
if n 10
is large, the minimum of E, l 2
and E# will be nearly equal
to Eir. Hence, we obtain an approximation to the minimum of Qi
by taking the minimum of the i numbers E* , Eji,...,E! i_ .
The quantity Eir can be computed as follows. Let g r be the
positive root in q of the equation
-41-
r+l a n-r-1 a
+ r+l+j =1-a
j=l jti- 1 ) r(r+l)+rj j(j+l)
Al 2 q] X1 2 2 qr+l+j (66)
(r6
Then
r(r+l ) (i-r)(i-r-)
2 i
E*2 r 1 2 gr
(67)
MINIMUM OF Q
0 0
Let ql,...,q n be values of ql,...,q n for which Qn becomes a
minimum. We shall prove that qj = (j =
0lqj ,...,n-).
oXq o
> Xq0
Assume that there exists a value j < n such that qj0
and we shall derive a contradiction. Let u be the smallest

integer such that qu+l > Xlqu and let v be the largest integer
0
suhthtq0 > 0 0 '
such that > lq Let qu= ( + C)qu (c > 0),
and q' = q for j p u, 4 v+l. Then for sufficiently

) )
small c we shall have Xlqr< qr+l < 2 q r r =
Furthermore, we have
ql ... qn n a.
ql ... qnjand j
= 1'"
q ... qo
Hence, there exists a positive A < 1 such that q7 q!

(j = l,...,n) and
n a.
E --- w =I - a0 4
j., q . q-
-42-
IT-
But then q" ... q < q~
0. q 0in contradiction to the assumption
1 n 1 n
that q° ... qn is a minimum. Hence, our statement is proved.
If q is the root of the equation

n a.
I a
S2 q]
n(n-l)
X1 2 qn
then the minimum of Qn is equal to
MAXIMUM OF Qi (i < n)
Lot q*,...,q* be values of q for which 0. becomes a

I n
maximum. We shall prove the following:
Lemma 3: The relations
q = Xq* i,...,n-.) (8
must hold.
Proof: Assume that there exists an integer j > i such that

q*+ > Alqj and we shall derive a contradiction. Let qrI = qr
for r and let q! X q (j = i,...,n-l). Then
n a.
q "'" q! = q .. q! and F q ..--- ] > I- a.
j=1
Hence, there exists a value A > 1 such that
n a.
jEl q1 ... q
where q! - Aq! (j = 1,...,n). But then qj ... q' > q ... ql in

contradiction to the assumption that q! ... q* is a maximum.
-43-
I'
Lemma 4: If for some j < i we have qj > then
q*+l= Xq for k = l,...,j-l.
Proof: Assume that qj > X qj for some j < i and that there
j+1 1 j
exists an integer k < j-1 such that q* < A2q " We shall
k+l 2k
derive a contradiction from this assumption. Let u be the
smallest integer such that q*
U+l < 2q* Furthermore, let v be
2u .
the smallest integer greater than or equal to u + 1 such that
u
qv > Xlq*v It is clear that v < j. Let q 1 C > 0),
q (1 + c) q, and q' q* for r yi u, / v. Then for suffi-
v v r r
ciently small c we have
lq' < J < L2q! (j = i, ... n-l)
n a.
=
qj... q! q ... ql and qq' > 1- a
Hence, there exists a value A > 1 such that

n a.
E -u--I--= 1 -a
lq" ... q!
where q7 = Aq! (j = i,...,n). But then q" ... q! ".

contradiction to the assumption that q* ... qt is a maximum.
(r a l,...,i-l)
Let Dir irq be the maximum of Q.
1 under the restric-
tion that q j+l = 1q.j for j = r+l,...,n-I and qj+l = X2q. for
j = l,...,r-l. From Lemma 3 and 4 it follows that the maximum of
Q. is equal to the maximum of the i - 1 values Dil,...,D _
ii
The computation of the exact value of D.i can be carried out in a
way similar to the computation of Mir in part II. Since these
computations are involved if n is large, we shall discuss here
only an approximation method.
-44-
Let D* (r = l,...,i-l) be the value of Q. if q = Alq for
* 1 j+1 1)i
j = r+l,...,n-l and qj+ X2 qj for j - l,...,r. Furthermore,
let D* be the value of Q. if qj+l = Xlqj (j = l,...,n-l). Then,
10and
if A is not much below one, the maximum of D and D#
I
(r = l,...,i-l) will be nearly equal to Dir. ir i,r-l
Hence, we obtain an
approximation to the maximum value of i by taking the largest of
the i values D ...,D*
10 1,i-I"
The value of D can be determined as follows. Let g be the

ir r
root in q of the equation
r+l a n-r-l ar+l+j a

j=l r(r+l) + . j(j+l)
A22 qjX 2 A1 2 rq~
2 2J
Then
r(r+l) (i-r-l) (i-r)
D =2 2 + r 2 1
ir 2 gr
MAXIMUM OF On
We shall prove that the maximum of Qn is reached when qj+l= A2 q j

Denote by q* ... q* the values of q .. qn
for which Qn becomes a maximum. We shall assume that there
exists a value j < n such that qj 1 < Aqt and we shall derive
a contradiction from this assumption. Let u be the smallest and
v be the largest integer such that q* < X q* and q*+l < q "
u+l 2 u q < A2
q*
Let qu (E > 0), qv+l = (I + c) q*,, and qr = q; for
L 'qC or
r / u, d v+l. Then for sufficiently small e we shall have
lqr'-- qr+l '2 q r (r -
-45-
n q.
S ' = q* and > 1- a
j= q1 ... q 0
Hence, there exists a value A > I such that qj= Aq

and
(j = l,...,n)
n a
F= q . a0
j=l 1j**
u * in contradiction to the assumption

t tn n
that qn q is a maximum. Hence, our statement is proved.
The maximum of Q is equal to

n
n(n-l)
2 n
A q
2
where q is the root of the equation
n a.
j=l J (3- ) 0
A2 qJ
X2
NUMERICAL EXAMPLE
The same notation will be used as in the previous numerical

examples. The assumption of no sampling error, which is common
to all the previous examples, is retained. In part I it was
assumed that the qi, the probability of a plane surviving the
i-th hit, knowing that the first i - I hits did not down the
plane, were equal for all i (ql = q = " = qn = qo (say)).
Under this assumption, the exact value of the probability of a
plane surviving i hits is given by
i
Qi = go
In part III it was assumed that q > q2 qn Since no

lower limit is assumed in the decrease from qi to q i+1 only a
-46-
lower bound to the Q i could be obtained. The assumption here is
that the decrease from qi to qi+l lies between definite limits.
Therefore, both an upper and lower bound for the can be
obtained.
We assume that
Alqi -qi+l 2 qi
where A1 < A2 < I and such that the expression
n a.
j~l<9jl) <1 a (A)
i 2
is satisfied.
The exact solution is tedious but close approximations to the

upper and lower bounds to the Qi for i < n can be obtained by
the following procedure. The set of hypothetical data used is
ao = .780 a 3 = .010
a 1 = .070 a 4 = .005
a 2 = .040 a 5 = .005
1 = .80 2 -. 90
Condition A is satisfied, since by substitution
.07 • + .04
-- t .83
.01 .0056 10
.005 "209
.20529
.8 (.8) (.8) (8)
which is less than
1 -a = .22
THE LOWER LIMIT OF Qi
The first step is to solve equation 66. This involves the

solution of the following four equations for positive roots go
g 1 ' g 2 ' g3 "
-47-
-~ - -~ - -~-~ -. ~
a1 a2 3 a4 a 5
+
~+ l--* 3 + +6 1I05'-' = 1 - a o 2 B
q X2 q X2q X2 q
.07 .04 .01 + .005 .005 5 .22

q .9q .729q .531441q .348678q
.22q 5 - .07q 4 - .044444q - .013717q 2 - .009408q - .014340 = 0
= 844.
a1 a2 a3 a4 a5
+ 2 3 .33 4 + 4 6 5 ===1aI
2 3 a0 C
(C)
q
q xlq X1 X2 q 1 2
xA 1Xx 2q
.07 q + .042 + .6)9q3

.01 +
+ .005 )q4 + .005 )5
=.22
q .8q (.64)(.9)q (.512)(.729 (.4096)(.531441)q
.22q - .07q - .05q - .017361q - .013396q - .022970 = 0
g = .904.
a. a2 a3 a4 a5
q X+q 2 X3 q 3 X5 xq4+ 7 3 = -a
Xq 2 3q3 2 X2
.07 + .04 + .01 + .005 + .005 .22

5
q .8q2 .512q 3 (.32768)(.9)q 4 (.209715)(.729)q
5
2 407q2
.22q 5 _07q 4 .05q - .019531q - .016954q - .032705 = 0
2= .941.
-48-
a, a2 a3 a4 + a5a(E
-+ + 3 -~ -3- 5 =1-a 0 (E
q AIq Ai~q Aq Al A2 q
.07 + .04 + .01 + .005 + .005 =.22

2+ 4
5
2 3 .262144q (.134218)(.9)q
q Sq .512q
.22q5 - .07q 4 .05q3 _ .019531q 2- .019073q -. 041392 -0
9 3 = .964
Next, calculate the i numbers defined by
E = ,a(ilr) Ab(ilr) g'ir (r

Ir 1 2
where
a(i,r) r(r 2 + 1) + r(i -r -1)
b~ir)= r)(i -r - 1)
= .844
= .904
92= .941
93= .964
The minimum of the V~ 0,...,i-1) will be the lower limit of

(r
Q.. The computations are given in table 5.
-49-
TABLE 5
COMPUTATION OF LOWER LIMIT OF
Q. i r a(i,r) b(i~r) g ri
E
I1
1 0 0 0 .844 .844 .844
Mini [E~ 0 .844
Q2 2 0 0 1 .844 .712 .641
2 1 1 j 0 .904 .817 .654
Min [E*0 , E*1 .641
Q3 0 0 3 .844 .601 .438
3 1 2 1 .904 .739 .426
3 2 3 0 .941 .833 .427
Min [E3 0 1 E~1 , E*2 .426
Q4
3 4 3 0 320 6 .844 .507 .270
4 1 3 3 .904 .668 .249
4 2 5 1 .941 .784 .231
4 3 6 0 .964 .864 .226
Mini [E~I E* 1 , E*2 E* 3 .226
-50-
The lower limit of can be obtained directly. The lower limit
of
where q is the positive root of
a a a a
2
1 + a 4. a
3
+ ~ 4
+
0 5
.07 1..04 + .01 + .005 + .005 =.22

2 3 4 5
q .8q .1q .262144q .107374q
q = .974
The lower limit of
Q5 = (.8) 10 (.974) 5 = 094
THE UPPER LIMIT OF 0.i
The computations for the upper limit of Q1are entirely analogous

to the computations of the lower limit. First, we solve the
equations of part IV, which for this example are the following:
2 33 050
.07 .04 + .01 + .005 + .005 .2

-723 4 5in2
q .8q& .51.2q .262144q .107374q
.22q 5 - 07q 4 - 05q 3 - 019531q 2 - 019073q -. 046566 -0
g* .974
-51
al a2 + a3 + a4 + a5 =-
2 2 3 3 34 4 65 0
q q xxq x Aq A l
.07 + .04 + .01 -+ .005 +.005 .22
5 2 3 2
.22q5 .07q 4 .044444q 3 .015432q2 _ .013396q - .029071 =0
=* .905
al a2 a3 a4 a5
q ~ \2qq 2 l \xq 0o
.07 +.04 + .01 .005 .005.2

q .92 .729q (.59049)(.8)q4 (.512)(.478297)q 5 2
.22q 5 .07q 4 .044444q 3- .013717q 2 .010584q -. 020417 =0
g*= .869
al a2 a3 a4 a5
2+ 33
-4 - = 1- a
q 2qXq 0
.07 .04 .01 .005 -+.005 - 2

2 3 4 2
q .9q2 .729q3 .531441q4 (.387420)(.8)q5
.22q 5- .07 q 4 .044444q 3 .013 7 17q 2 .009408q -. 016132 =0
-* .851
-52-
Next, calculate the i numbers defined by
b(i,r)
a(i,r) 19(r *i i-l),
Dtr )2 -0,1...
where
a(i,r) r(r 2+ 1) + r(i -r 1)
g .974
9*- .905
9*- .869
9*= .851
The maximum of the D (r 0,...,i-1) will be the upper limit of

Q* The computations are given in table 6.
The upper limit of 05can be obtained directly. The limit of
X10 *5
5 ~2 q
where q* is the positive root of
a1 a 5
q ,2q A2 q X2q A2 q
.07+
.4 .0 + .005 + .005 - .22
q .9q2 729q3 531441q4 348678q 5
-* .844.
-53-
TABLE 6 F
COMPUTATION OF UPPER LIMIT OF 0.
Qi r a(i~r) b(i~r) g qrW
01 1 0 0 0.7 94 .7
Max ED* 0 1 .974
Q2 2 o0 0 1 .974 .949 1 75
2 1j 1 0 .905 .819__ .737
Max tD~0 , D*1 .759
Q3 3 0 0 3 .974 .924 .473
3 1 2 1 .905 .741 .480

3 2 3 0 .869 .656 .478
Max [DSOf D~1 , D*2 .480
Q4 14 4
0
1
0
3
6
3
.974
.905
.890
.671
.236
.250
4 2 5 1 .869 .570 .269
4 3 6 0 .851 .524 .279
Max [D~0 I D~1 , D*2 D*3 .279
-54-
The upper limit of
Q-(.9)1 (.844)~
-149
Summarizing the results, the upper and lower limits of the

probability of a plane surviving i hits are given by
.844 < < .97

.641 < 02 4 .759
.426 < 0 ' .480
.226 < 0 ' .279
.094 < ( .149
PART V
SUBDIVISION OF THE PLANE INTO SEVERAL

EQUI-VULNERABILITY AREAS 1
In parts I through IV we have considered the probability that a

plane will be downed by a hit without any reference to the part
of the plane that receives the hit. Undoubtedly, the probability
of downing a plane by a hit will depend considerably upon the
part that receives the hit. The purpose of this memorandum is to
extend the previous results to the more general case where the
probability of downing a plane by a hit depends on the part of
the plane sustaining the hit. To carry out this generalization
of the theory, we shall subdivide the plane into k equi-
vulnerability areas AI,...,Ak. For any set of non-negative
integers ilI...,k let P(il,...,ik) be the probability that a
plane will be downed if the area A receives i1 hits, the area
A receives i hits,..., and the area A receives i hits. Let
2 2 k k
Q(il,...,ik) = 1 - P(il,...,ik). Then Q(il,...,ik) is the prob-
ability that the plane will not be downed if the areas Al,...,Ak
receive i ,...,i k hits, respectively. We shall assume that
Q(il,...,ik) is a symmetric function of the arguments il,...,. k .
To estimate the value of Q(il,...,ik) from the damage to

returning planes, we need to know the probability distribution of
hits over the k areas A,,...,Ak knowing merely the total number
of hits received. In other words, for any positive integer i we
need to know the conditional probability Y1(i,...,ik) that the
areas Al,...,Ak will receive il,...,ik hits, respectively,
knowing that the total number of hits is i. Of course,
yi(il,...,ik) is defined only for values il,...,ik for which

iI + ... + ik = i . To avoid confusion, it should be emphasized
that the probability Yi(il,...,ik) is determined under the

AMP memo 76.5.
-56-
I.
assumption that dummy bullets are used. It can easily be shown
that it is impossible to estimate both Yi(il,...,ik) and
Q(ilb...,ik) from the damage to returning planes only. To see
this, assume that k is equal to 2 and all hits on the returning
planes were located in the area A.. This fact could be explained
in two different ways. One explanation could be that
Y(i,i2) =0 for i2 > 0. The other possible explanation would be that
Q(ili 2 ) = 0 for i2 > 0. Hence, it is impossible to estimate
both yi(il,i 2 ) and Q(i1 1 i 2 ). Fortunately, Yi(il,...,ik) can be
assumed to be known a priori (on the basis of the dispersion of
the guns), or can be established experimentally by firing with
dummy bullets and recording the hits scored. Thus, in what
follows we shall assume that Yi(il,...,ik) is known for any set
of integers il.,k
Clearly, the probability that i hits will not down the plane is
given by
Qi Yi "I ,(i Y .. Y (69)

Ik 1
where the summation is to be taken over all non-negative integers

il,...,i k for which i I + + ik =
Let 6i (i ,...,ik) be the conditional probability that the areas

A,...,Ak received il,...,ik hits, respectively, knowing that the
plane received i hits and that the plane was not downed. Then we
have
y (i,...,ik)Qlil,...,ik)
6i(il,...,ik) -a 0 (70)
Of course, 6i(il,....ik) is defined only for non-negative

integers i i...,.
k for which i + ... + ik - i
-57-
O
-N NE~ N
6
The probability i(il ... ik) can be determined from the distri-
bution of hits on returning planes. In fact, let a(il,...8ik)
be the proportion of planes (out of the total number of planes
participating in combat) that returned with i hits on area Al,
i 2 hits on area A 2 ,.... and ik hits on area Ak. Then we
obviously have
i i . .k) a.1 (71)
From equations 70 and 71, we obtain
Qia(il , ...,ik)
Si l k 1 (72)
Since Qi can be estimated by methods described in parts I through

IV, estimates of Q(il,...,ik) can be obtained from equation 72.
According to equation 29, the probabilities Ql ..... Qn satisfy the

equation
n a.
E -2 = 1 - a0 (73)
j=1 Qj
-
We have assumed that q q2 "'" >- q n' This is equivalent to
stating that
Qi+l 4 Qj+L for j < i . (74)

Qi - Qj
A similar assumption can be made with respect to the prob-

abilities Q(il,...,ik). In fact, the conditional probability
that an additional hit on the area A
r will not down the plane
knowing that the areas A1 ,...,A k have already sustained
il...,ik hits, respectively, is given by
-58-
........ M,
Q(i..,rl, r ...,
0") (75)
Obviously, we can assume that if
Ji-S ill J2 i i2"''J k S ik
then
Q•'* " r
Q(Ierir,Ir iLr+" , li,...-,li~
k) < 01 * oj -'r+lJr+l'*~ k
I , -(l.-J~lJ,.,k
Q(Tl..I r-l' r r+l,...,ik) - Q(Jl,...,Jr-l,3rJr+l, ...,Ilk

)
(76)
for r = l,2,...,k.
Hence, the possible values of Ql.'.Qn are restricted to those

for which equation 73 is fulfilled and for which the quantities
Q(il,...,ik) computed from equation 72 are less than or equal to
one and satisfy the inequalities of equation 76. It should be
remarked that the inequalities of equation 76 do not follow from
the inequalities of equation 74. From equation 72 and the
inequality Q(il,...,ik) 11, it follows that
aT (il,.-.,ik)
0Q < (77)
i - a(il,...,ik)
If the right-hand side expression in equation 77 happens to be

less than one, then equation 77 imposes a restriction on Qi"
Since
a(il,..., k )
k 1 k I1
(the summation is taken over all values il,...,ik for which

*
if+ ... + ik = i), we must have either
-59-
for all values il,...,ik for which iI + ... + ik = i, or
aiYi i , •.., k ) <

a(il,...,ik) <1
at least for one set of values il,...i"k satisfying the condition

iI + ... + ik = i . Hence, equation 77 gives an upper bound for
Qi whenever there exists a set of integers il,...,k such that
i ... + i = i and
1 k
a(la •i ( l...iF
It is of interest to investigate the case of independence, i.e.,

the case when the probability that an additional hit will not
down the plane does not depend on the number and distribution of
hits already received. Denote by q(i) the probability that a
single hit on the area A. will not down the plane. Then under
1
the assumption of independence we have
2 1
Q(il,...,ik) = [q(l)J [q(2)] ... q(k) . 78)
Hence, the only unknown probabilities are q(l),...,q(k).

Let Y(i) be the conditional probability that the area A i is hit
knowing that the plane received exactly one hit. Obviously
i i
[Y(1)] i .. [y(k)) (79)
Yi i Yi(l' k
,.,k) 1 1 . iki
-60-
- II
4 aR
Similarly, let 6(i) be the conditional probability that the area

A. is hit knowing that the plane received exactly one hit and
1
this hit did not down the plane. Because of the assumption of
independence, we have
I' k
6i'l,...,k ) i1 ! ... k! [6(1)1 ...* [6(k)] .(G0)
6(i) =- k
y(i)q(i)
~ Y(i)q(i)
i=I
Since the probability q that a single hit does not down the plane
k
is equal to . Y(i)q(i), we obtain from equation 81
Y(i)
q(i) =q (82)
Because of the assumption of independence, we see that 6(i) is

equal to the ratio of the total number of hits in the area A of
I
the returning planes to the total number of hits recei.. by the
returning planes. That is
F ...F Jia(Jl,--.',Jk )
Jk 31
,3
6(i) = ~ l.(3
+ +
"'" Jl "" k)a(jl''''. k
S(j 1 +
*** +j
3k 31
Since Y(i) is assumed to be known and since 6(i) can be computed

from equation 83, we see from equation 82 that q(i) can be
determined as soon as the value of q is known. The value of q
can be obtained by solving the equation
j=l a- (84)
-61-
NUMERICAL EXAMPLE
In the examples for parts I, III, and IV we have estimated the

probability that a plane will be downed without reference to the
part of the plane that receives the hit. However, the vulner-
ability of a particular part (say the motors) may be of interest
and this example illustrates the methods of estimating part
vulnerabilities under the following assumptions:
0 The number of planes participating in combat is large so

that sampling errors can be neglected.
* The probability that a hit will down the plane does not
depend on
= the number of= previous non-destructive hits. That
is, ql q2 "' qn qo
o Given that a shot has hit the plane, the probability that
it hit a particular part is assumed to be known. In this
example it is put equal to the ratio of the area of this
1
part to the total surface area of the plane.
* The division of the plane into several parts is repre-

sentative of all the planes of the mission. If the types of
planes are radically different so that no representative
division is possible, we may consider the different classes
of planes separately.
Consider the following example. Of 400 planes on a bombing

mission, 359 return. Of these, 240 were not hit, 68 had one hit,
29 had two hits, 12 had three hits, and 10 had four hits.
Following the example in part I we have
N = 400,
whence
A = 240 a = .600
0 0
A = 68 a = .170
1 1.
A = 29 a = 072
22
A3 = 12 a3 = 030
A4 = 10 a 4 =. 025
lBy area is meant here the component of the area perpendic-

ular to the direction of the enemy attack. If this direction
varies during the combat, some proper average direction may be
taken.
-62-
As before, the probability that a single hit will not down the
plane is given by the root of
a2 3 4
q qo qo qo
which reduces to
°
.4q - - 0 72q .030q _ .025 0
0
and
qo= .850.
Suppose that we are interested in estimating the vulnerability of

the engines, the fuselage, and the fuel system. Assume that the
following data is representative of all the planes of the
mission:
Ratio of
area of part
to total
Part number Description Area of part area (Y(i))
1 2 engines 35 sq. ft. -= .269
2 Fuselage 45 sq. ft. = .346
3 Fuel system 20 sq. ft. 20 .154

130
4 All other parts 30 sq. ft. 3 = .231
Total area 130 sq. ft.
The ratio of the area of the i-th part to the total area is
designated Y(i). Given that the plane is hit, by the third
assumption, Y(i) is the probability that this hit occurred on
part i. Thus
-63-
Y(l) = .269
Y(2) = . 346
Y(3) = .154
Y(4) = .231
The only additional information we require is the number of hits

on each part. Let the observed number of hits be 202. In
general, the total number of hits (on returning planes) must be
equal to
A i + 2A 2 + 3A3 + ... + nA
and in this example
A + 2A + 3A 3 + 4A = 68 + 2(29) + 3(12) + 4(10) = 202
The hits on the returning planes were distributed as follows:
Ratio of number of hits

observed on part to
Number of hits total number of observed
Part number observed on part hits (6(i))
1 39 .193
2 78 .386
3 31 .154
4 54 .267
Total number of hits 202
The ratio of the number of hits on part i to the total number of

hits on surviving planes is designated 6(i). Then q(i), the
probability that a hit on the i-th part does not down the plane,
is given by
q(i) = qo
-64-
- .'
t.
" .
. -
whence
6(1) .193
q(l) = f-(1 q0 .269 .850) 61
6(2) .386 95
q(2) Y C2 So .346 .850)
YT3) =
q(3) 6(3) ' o .154"
.154 (.850) = 85
q(4) 6(4) .267 .5) 9

q(4) = __ qo = .231 (.850) = .98
The results may be summarized as follows:
Probability of Probability of being

surviving a single downed by a single
Part hit (g(i)) hit (1 - g(i))
Entire plane .85 .15
Engines .61 .39
Fuselage .95 .05
Fuel system .85 .15
Other parts .98 .02
Thus, for the observed data of this hypothetical example, the

engine area is the most vulnerable in the sense that a hit there
is most likely to down the plane. The fuselage has a relatively
low vulnerability.
- 5
-65-
a"
.w - .. ..
.... .. . . . . .. ..... . ..
........ ........ .___
PART VI
SAMPLING ERRORS 1
In parts I through V we have assumed that the total number of

planes participating in combat is so large that sampling errors
can be neglected altogether. However, in practice N is not
excessively large and therefore it is desirable to take sampling
errors into account. We shall deal here with the case when
ql = q2"'" = qn = q (say) and we shall derive confidence limits tor
r
the unknown probability q.
If there were no sampling errors, then we would have

(b5)
1 = p(l - a -a a - x - x - " i-I
(i = 2,3,...),
where p = 1 - cj. However, because of sampling errors we shall

have the equation
x. = p (l-a - ...- a -x - ... - xil (L6)
where pi is distributed like the success ratio in a sequence of

Ni = M1( - a - a1 - ... - a i-I - - ... - x i_) independent
trials, the probability of success in a single trial being equal

to p.
Let qi = I - pi " Then, according to equation 26 we have
n a.
I - a0 , (07)
j=l q1 ... q

AMP memo 76.6.
-66-
provided that x. = 0 for i > n. In part I we have shown that
x i = 0 for i > n if there are no sampling errors. This is not
necessarily true if sampling errors are taken into account. However,
in the case of independence, i.e., when qi = q (i = 1,2,...), x i
is very small for i > n so that x. can be neglected.
In fact, if the number of planes that received more than n hits

were not negligibly small, it follows from the assumption of
independence that the probability is very high that at least some
of these planes would return. Since no plane returned with more
=0
than n hits, for practical purposes we may assume that x i=.
In what follows we shall make this assumption. i=n+l
Each of the quantities ql,...,q n can be considered as a sample

estimate of the unknown probability q. However, the quantities
ql,...,q n are unknown. It is merely known that they satisfy
the relation in equation 87. Confidence limits for q may be
derived on the basis of equation 87. However, we shall use
another more direct approach.
To derive confidence limits for the unknown probability q we shall

consider the hypothetical proportion b. of planes that would have
1
been hit exactly i times if dummy bullets would have been used.
We shall treat the quantities bl,...,bk as fixed (but unknown)
constants. This assumption does not involve any loss of
generality, since the confidence limits for q obtained on the
basis of this assumption remain valid also when b1,...,b k are
random variables. Clearly, the probability distribution of Na1
(i = l,...,n) is the same as the distribution of the number of
successes in a sequence of tib i independent trials, the prob-
i
ability of success in a single trial being q . Hence
E(Nai) = q Nbi (W)
*2 i( i
a (a.) = Nbiq 1 - q ) . (e9)
-67-
From equations 88 and 89 we obtain
E (a) =bi (90)
(91)
a2 b (1-q'
a1 a 2 an
Since the variates -, 2""' are independently distributed,
q q q
and since ai is nearly normally distributed if N is not small, we
can assume with very good approximation that the sum
n a.
(92)
i=1 q
is normally distributed. We obtain from equations 90 and 91
E ) = E bi =1 - a (93)
) i=l b
(i~l 1=1 Nq'
For any positive a < 1 let X. be the value for which

2
A ~t
S1 e 2 dt = a
The set of all values q for which the inequality
i-a --
i b(-q n
a
~b1
+k
f
q -- -tq' F (95)
-68-
ft
is fulfilled forms a confidence set for the unknown probability q
with confidence coefficient a . However, formula 95 cannot be
used, since it involves the unknown quantities b,...,bn . Since
aa .
-r converges stochastically to b. as N--, we change the stan-
q a. I i-
dard deviation of E, only by a quantity of order less than
qi
it we replace b. by . Thus, the set of values q that satisty
the inequalities
-a - I - n --- i - a ° +( - q (96)
1z~
-- - s i - o a 2Tq1-
is an approximation to a confidence set with confidence

coefficient a
Denote by q0 the root of the equation in q
r') a.
q]0
*j q) o
Then q converqes stochastically to q as N . A considerable

simplitication can be achieved in the computation of the
confidence set by substituting q0 for q in the expression ot the
a.
standard deviation of 1-j. . The error introduced by this substi-
q
tution is small if 1j is large. Making this substitution, the
iirequalities uctining the confidence set are given by
1I - ao -'( j - n a) n al - 'q
-69-
-,.•. . . . .. . " : . . . , . . . ... . . ._ _ _

Hence, the confidence set is an interval. The upper end point of
the confidence interval is the root of the equation
O )
=- I - a0 -A a Y
i=l q Nq()
and the lower end point of the confidence interval is the root ot
the equation
n a n a (1-i
=1 a o 2i_
i=l q 0 +) i=l Nq - (o9
NUMERICAL EXAMPLL
In all previous examples it was assumed that A. (the number oL

1
jlanes returning with i hit;) was compiled from such a large
number of observations that they were not subject to sampling
errors. If it is further assumed that the probability q that a
hit will down a plane does not depend on the number of previous
non-destructive hits, it is possible to obtain an exact solution
for the probability that a hit will down a plane. Here we
introduce the possibility that the A ,...,A are suLject to
0 n
sampling errors but retain the assumption of independence. Lndur
these less restrictive assumptions we cannot obtain the exaect
solution for q, but for any positive number a < I we can construct
two functions of the data, called confidence limits, such that
the statement that q lies between the confidence limits will be
true 100a percent of the time in the long run. The contidence
limits are calculated for a = .95 and .99.
-70-
mI
It
Under the assumptions of part I, it was proved that no planes

received more hits than the greatest number of hits observed on a
returning plane. This is not necessarily true when the possi-
bility of sampling error is introduced, but it is retained as an
assumption, since the error involved is small.
If the a. are subject to sampling error, and q is the true para-

1
meter,
n a.
1 (A)
1 q
will be approximately normally distributed with mean value 1 - a0
In outlining the steps necessary to calculate the confidence

limits, the following hypothetical set of data will be used.
Given
Ai
N =500 ai = Fr
A =400 a = .80
o o
A1 40 a I = .08
A = 25 a 2 = .05
A = 5 a 3 = .01
A = 3 a 4 = .006
A = 2 a 5 = .004
475
The first step is to find the value q0 for which expression A is

equal to its mean value, by finding the positive root of
a a2 a3 a4 a5
-I + - - + = I - ao
ci q2 q3 q4 q5
We obtain
.20q 5 2
-. 08q 4 - .05q 3 - .0q - .O06q - .004 = 0
qo =850.
-71-
.~.. ~
Wit~'%S s~r .,J.fl. .a~t 4g.. t., ~ _
The next step is to calculate the standard deviation of
expression A. This can be shown to be approximately equal to
l 2 "q
S2 3 4
1 a(1-q0 ) a 2 (l-qo) a 3 (l-q O) a 4 (1-q 0 a 5 (-q 0 )
2
-- + + + 8 + 0-
°
Nq Nq Nq Nq Nqo
0 0 0 0 0
= .01226
n a.
Knowing that I is approximately normally distributed with
i1l q
mean value 1 - a0 and the standard deviation 0, we can determine
the range in which Z can be expected to be 100a percent of

q
the time (say 95 and 99 percent) by determining X 95 and .99
such that
1 .95 exp 2 dt - .95
f 2.95
-X .99
From the table or the areas of a normal curve, it is found that
A .95 = 1.959964
.9 9 = 2.575829
We can now calculate the confidence limits for each value of a by

finding the two values of q for which the equality sign of the
following expression holds:
-72-
i -L- (1- a <
i 0a) < a O
It follows that for each a, the confidence limits are the

positive roots of the equation
n a.
i --L -a +x
- 3Z .0122678X 1 - a - a I - ° +X y
-a-_ a__ 1_0 a 0 aI
.95 1.959964 .024044 .175956 .224044
.99 2.575829 .031600 .168400 .231600
For a = .95 the confidence limits of qo are the positive roots o

equat ior.
22-+ ~q3
- +. 4 4 +175956,
5
which reduces to
5 4 3 2
.175956q - .08I - .05q - .1q 2 - .OU6q- .004 = U
= .912,
and equation
aI a., a3 ri a
S- + + - + - + -= .224044,
c, 3 '14 S5
t which reduces to
.224044q 5 - .Obq 4 - .U5q 3 - .Olq 2 - .OObq - .004 = 0

(1 = •.801. k
Similarly, tor a = .99 we have
-73-
3 -_.Olq 2
.168400q 5 - *08q4 - .05q _ .006q - .004 = 0
q = .935
.231600q 5- .O08q 4- .05q 3 _.Olq 2- .006q - .004 = 0

q = .787
Summarizing the results we find that the 95-percent confidence

limits of q are .801 and .912, and that the 99-percent confidence
limits are .787 and .935.
-74-
PART VII
MISCELLANEOUS REMARKS 1
1. Factors that may vary from combat to combat but influence the
probability of surviving a hit. The factors that influence the
probability of surviving a hit may be classified into two groups.
The first group contains those factors that do not vary from
combat to combat. This does not necessarily mean that the factor
in question has a fixed value of all combats; the factor may be a
random variable whose probability distribution does not vary from
combat to combat. The second group comprises those factors whose
probability distribution cannot be assumed to be the same for all
combats. To make predictions as to the proportions of planes
that will be downed in future combats, it is necessary to study
the dependence of the probability q of surviving a hit on the
factors in the second group. In part V we have already taken
into account such a factor. In part V we have considered a
subdivision of the plane into several equi-vulnerability areas
A 1 ,...,Ak and we expressed the probability of survival as a func-
tion of the part of the plane that received the hit. Since the
probability of hitting a certain part of the plane depends on the
angle of attack, this probability may vary from combat to combat.
Thus, it is desirable to study the dependence of the probability
of survival on the part of the plane that received the hit. In
addition to the factors represented by the different parts of the
plane, there may also be other factors, such as the type of gun
used by the enemy, etc., which belong to the second group. There
are no theoretical difficulties whatsoever in extending the
theory in part V to any number and type of factors. To
illustrate this, let us assume that the factors to be taken into
account are the different pirts Al,...,Ak of the plane and the
different guns g 1 ,...,g m used by the enemy. Let q(i,j) be the
probability of surviving a hit on part A 1 knowing that the bullet
has been fired by gun gj. We may order the km pairs (i,j) in a
sequence. We shall denote q(i,j) by q(u) if the pair (i,j) is
the u-th element in the ordered sequence of pairs. The problem
of determining the unknown probabilities q(u) (u = l,...,km) can
be treated in exactly the same way as the problem discussed in

AMP memo 76.7.
-75-
part V assuming that the plane consists of km parts. Any hit on
part A. by a bullet from gun g] can be considered as a hit on
part Au in the problem discussed in part V where (i,j) is the
u-th element in the ordered sequence of pairs.
2. Non-probabilistic interpretation of the results. It is
interesting to note that a purely arithmetic interpretation of
the results of parts I through V can be given. Instead of
defininq qi as the probability of surviving the i-th hit knowinq
that the previous i - I hits did not down the plane, we define qi
as follows: Let M i be the number of planes that received at least
i hits and the i-th hit did not down the plane, and let I].
1 be the
total number of planes that received at least i hits. Then
M1
= -. "Thus, qi is defined in terms of what actually hap-
pened in the particular combat under consideration. To oistin-

guisn this definition of qi from the probabilistic definition, we
M.
shall denote the ratio L
-- by j". 'ihe quantity (] is unknown,
1
since we do not know the distribution of hits on the planes tihat
did not return. However, it follows from the results of part 1
that these quantities must satisfy equation 26. If we can assume
that in theparticular combat under consideration we have
qi = ... = qn then the common value q o£ these quantities is the
root of the equation
a.
-a 0
> >
Assuming that ql q 2 > "" n ' the minimum value Oi .i
derived in parts III and IV can be interpreted as the minimu.i,

value of Qi = ql ... qi"
The minimum and maximum values of Q. derived in part IV can also

1
be interpreted as minimum and maximum values of Q.= q I"" ' if
we assume that the inequalities XIqj S qj+l < 2j = l,...,n-l)
are fulfilled. Similarly, a pure arithmetic interpretation of
the results of part V can be given.
-76-
- - - -
3. The case when Y(i) is unknown. In part V we have assumed
that the probabilities y(l),...,y(k) are known. Since the
exposed areas of the different parts AlI...,A k depend on the
angle of attack, and since this angle may vary during the combat,
it may sometimes be difficult to estimate the probabilities
Y(1),...,Y(k). Thus, it may be of interest to investigate the
question whether any inference as to the probabilities
q(l),...,q(k) can be drawn when Y(l),...,Y(k) are entirely unknown.
We shall see that frequently a useful lower bound for q(i) can
still be obtained. In fact, the value q*(i) of q(i), calculated
under the assumption that the parts A.(j
KJ pi i) are not vulnerable
(q(j) = 1), is certainly a lower bound of the true value q(i).
Considerinq only the hits on part Air a lower bound of q*(i), and
therefore aiso of q(i), is given by the root of the equation
n a*
r =I-a*(I0
r 0 (100)
r7 q o
where a* (r = 0,1,...,n) is the ratio of the number of planes

r
returned with exactly r hits on part A i to the total number of
planes participating in combat.
The lower limit obtained from equation 100 will be a useful one
if it is not near zero. The root of equation 100 will be
n
considerably above zero if L a* is not very small as compared
r= r
with 1 - a*. This can be expected to happen whenever both Y(i)
0
and q(i) are considerably above zero.
-77
-7"-
PART VIII
VULNERABILITY OF A PLANE TO DIFFERENT TYPES OF GUNS 1
In part V we discussed the case where the plane is subdivided

into several equi-vulnerability areas (parts) and we dealt with °
the problem of determining the vulnerability of each of these
parts. It was pointed out in part VII that the method described
in part V can be applied to the more general problem of esti-
mating the probability q(i,j) that a plane will survive a hit on
part i caused by a bullet fired from gun j. However, this method
is based on the assumption that the value of Y(i,j) is known
where Y(i,j) is the conditional probability that part i is hit by
gun j knowing that a hit has been scored. In practice it may be
difficult to determine the value of y(i,j) since the proportions
in which the different guns are used by the enehy may be unknown.
On the other hand, it seems likely that frequently we shall be
able to estimate the conditional probability Y(ilj) that part i
is hit knowing that a hit has been scored by gun j. The purpose
of this memorandum is to investigate the question whether q(i,j)
can be estimated from the data assuming that merely the quan-
tities y(ilj) are known a priori. In what follows we shall
restrict ourselves to the case of independence, i.e., it will be
assumed that the probability of surviving a hit does not depend
on the non-destructive hits already received.
Let 6(i,j) be the conditional probability that part i is hit by
gun j knowing that a hit has been scored and the plane survived
the hit. Furthermore, let q be the probability that the plane
survives a hit (not knowing which part was hit and which gun
scored the hit). Then, similar to equation 82, we shall have
q(ij) q " (101)
Let q(j) be the probability that the plane will survive a hit by
gun j (not knowing the part hit). Then obviously
q(j) = YY (ilj)q(i,j) . (102)

1
Let 6(ilj) be the conditional probability that part i is hit by

gun j knowing that a hit has been scored by gun j and the plane
survived the hit. Clearly
'This part of "A Method of Estimating Plane Vulnerability

AMP memo 76.8.
-78-
6(ilj) = (iJj)q(i,j)
]y(i Ij )q(i,j) y(ilj)q(i,j)
q(j) (103)
From equation 103, we obtain
q(i,j) = y((i7j)
6(i1 q(J) "(104)
The quantity 6(ilj) can be estimated on the basis of the observed

hits on the returning planes. The best sample estimate of 6(ilj)
is the ratio of the number of hits scored by gun j on part i of
the returning planes to the total number of hits scored by gun j
on the returning planes. Thus, on the basis of equation 104, the
probability q(i,j) can be determined if q(j) is known.
Now we shall investigate the question whether q(j) can be

estimated. First, we shall consider the case when it is known a
priori that a certain part of the plane, say part 1, is not
vulnerable. Then q(i,j) = 1 and we obtain from equation 104
1 = 6(lL q(j) . (105)
Hence,
Y(ll
q(j) = (106)
Thus, in this case our problem is solved. If no part of the

plane can be assumed to be invulnerable, then we can still obtain
upper limits for q(j). In fact, since q(i,j) < 1, we obtain from
equation 104
q(j) < . (107)
Denote by p(j) the minimum of Yrij with respect to 1. Then

we have
q(j) < p(j) • (108)
If there is a part of the airplane that is only slightly

vulnerable (this is usually the case), then q(j) will not be much
below p(j). Let the part i. be the part of the plane least
-79-
vulnerable to gun j. If q(i.,j) has the same value for any gun
j, then q(j) is proportional to P(j). Thus, the error is perhaps
not serious if we assume that q(j) is proportional to p(j), i.e.,
p
q(j) = XP(j). (109)
The proportionality factor X can be determined as follows. From
equations 101 and 104 we obtain [
8(ij _______
y~~)q = AP(j) (i) (110)
Hence,
XY(i,j) = q 8k(i21P(i~ 11
Denote Z8(i,j) by 8(j). Then,
8(ilj) = 8j~
( . (112)
From equations 111 and 112 we obtain
XY(i,j) = q 6MY!i-j)~ (113)
Since
SY(ilj) = 1,
we obtain from equation 113
Aj Y(ij) q (114)
But
~ Y(ij) =1
Hence,
q 6((115)
Since 6(j) and p(j) are known quantities, the proportionality

factor X can be obtained from equation 115. The probability q is
the root of the equation
n a.
F, - = 1 - a0 ,
j=l qJ0
where a. denotes the ratio of the number of planes returned with

3
exactly j hits to the total number of planes participating in
combat.
NUMERICAL EXAMPLE
In part V, the case of a plane subdivided into several equi-

vulnerability areas was discussed, and the vulnerability of each
part was estimated. The same method can be extended to solve the
more general problem of estimating the probability that a plane
will survive a hit on part i caused by a bullet fired from gun j,
if assumptions corresponding to those of part V are made. The
first three of the four assumptions that must be made to apply
the method of part V directly are identical with those made in
part V. They are:
o The number of planes participating in combat is large so
that sampling errors can be neglected.
e The probability that a hit will not down the plane does
not depend on the number of previous non-destructive
hits. That is, q1 = q 2 = "'" = qo (say), where qi is the
conditional probability that the i-th hit will not down
the plane, knowing that the plane is hit.
o The division of the plane into several parts is
representative of all planes of the mission.
The fourth assumption necessary to apply the method of part V

directly usually cannot be fulfilled in practice. It is:
o Given that a shot has hit the plane, the probability that
it hit a particular part, and was fired from a particular
type of gun, is known.
-81-
These probabilities depend upon the proportions in which differ-
ent guns are used by the enemy. To overcome this difficulty a
method that does not depend on these proportions is developed in
part VIII. The assumptions necessary for the method of part VIII
differ from those of part V only in that the fourth assumption is
replaced by:
* Given that a shot has hit the plane, and given that it
was fired by a particular type of gun, the probability
that it hit a particular part is known.
The information necessary to satisfy this assumption is more

readily available, and in the numerical example that follows a
simplified method is suggested for estimating these
probabilities.
The Data
The numerical example will be an analysis of a set of hypotheti-
cal data, which is based on an assumed record of dhmage of sur-
viving planes of a mission of 1,000 planes dispatched to attack
an enemy objective. Of the 1,000 planes dispatched, 634 (N)
actually attacked the objective. Thirty-two planes were lost
(L=32) in combat and the number of hits on returning planes was:
A. - number of planes returning with i hits
A = 386
A = 120 (A)
1
A2 = 47
A3 = 22
A4 = 16
A = 11
The total number of hits on all returning planes is
A1 + 2A 2 + 3A 3 + 4A 4 + 5A =
5
(B)
120 + 2x47 + 3x22 + 4x16 + 5xll = 399
These 399 hits were made by three types of enemy ammunition:
-82-
B1 Flak
B 20-mm aircraft cannon
B 7.9-mm aircraft machine gun
and the hits by these different types of ammunition were also

recorded by part of airplane hit:
C1 Forward fuselage
C2 Engine
A C3 Full system
C4 Remainder
The necessary information from the record of damage is given in
table 7.
TABLE 7
NUMBER OF HITS OF VARIOUS TYPES BY PARTS
Forward Fuel tal

fuselage, Engine, system, Remainder, all
Cl C2 C3 C4 parts
Flak, B 1 17 25 50 202 294
20-mm 8 7 17 18 50
cannon, B 2
7.9-mm 7 13 17 18 55
machine
gun, B 3
Total all 32 45 84 238 399

types
A Method of Estimating the Probability of Hitting a Particular

Part Given That a Shot of a Particular Ammunition Has Hit the
Planel
The conditional probability that a plane will be hit on the i-th

area, knowing that the hit is of the j-th type, must be deter-
mined from other sources of information than the record of
lNecessary for fourth assumption.
-443-
damage. Although a simplified method is used in this example,
more accurate estimates can be made if more technical data is at
hand. The first step is to make definite boundaries for the
areas C1 , C2 , C3 , C4 . Next, assume that each type of enemy fire
B1 , B2 , B 3 has an average angle of fire 618 02, 03* Finally,
assume that the probability of hitting a part of the plane from a
given angle is equal to the ratio of the exposed area of that
part from the given angle to the total area exposed from that
angle.
In this example it is assumed that flak (B) has the average

angle of attack of 45 degrees in front of and below the plane,
whereas 20-mm cannon and 7.9-mm machine gun fire both hit the
plane head-on on the average. The area C 1 is so bounded that it
includes areas which, if hit, will endanger the pilot and
co-pilot. Area C 2 includes the engine area and area C 3 consists
essentially of the area covering the fuel tanks. The results of
computations, based on the above assumptions, are assumed to be
as follows, where Y(C lB.) represents the probability that a hit
is on part Ci knowing it is of type B. (as estimated by deter-
mining the ratio of the area of C i to the total area as viewed
from the angle 0. associated with ammunition B.).
(C)
Y(C 1 81 ) = .058 Y(C1 B 2 ) = .143 Y(C IB 3 ) = .143
Y(C 2 B I ) = .092 Y(C 2 1B2 ) = .248 7(C 2 B 3 ) = .248
Y(C 3 B ) = .174 Y(C3 1B2 ) = .303 Y(C3 B 3 ) = .303
Y(C4 1BI ) = .676 7(C 4 IB 2 ) = .306 Y(C4 B3 ) = .306
1 Thisnotation differs from the previous notation of part

VIII. In the first part of part VIII, Y(ilj) is used with the
understanding that the first subscript refers to the part hit and
the second subscript refers to the type of bullet. In the
numerical example, the relationship is made explicit by letting
C stand for the i-th part (or component) and B. for the j-th
i J
type of bullet. The same device is used throughout this example.
-84-
Computations for Method of Part VIII
Let q(Ci,B j) be the probability of surviving a hit on part C. by [
gun Bj. By equation 104, we have v
6(C. lB.
q(Ci,B j ) = (i-Bj) q(Bj) (D)
where b(C. IB.) is the probability of being hit on part C.,

knowing that the hit was scored by a bullet from gun B.) and that
the plane survived; 7(Ci B.) is the probability of being hit on
part C.i, knowing that the hit was scored by a bullet of type B.;
and q(B.) is the probability that a plane will survive a hit of
type B., knowing that the plane is hit. This can be estimated by
taking the ratio of the number of hits of type B.
J on part C.1 to
the total number of hits of type B. on returning planes.
3
Applying this method to the table we obtain
(E)
(CIB1) = .058 6(C l I B 2 ) = .160 6(C1 IB 3) = .127

6(C 2 1B1 ) = .085 6(C 2 1B 2 ) = .140 b(C 2 1B 3 ) = .236
&(C 31B 1 ) = .170 b(C 3lB 2 ) = .340 6(C 3 1B 3 ) = .309
s(C 4 1B I ) = .687 6(C 4 1B 2 ) = .360 6(C 4 1B 3 ) = .327
The final quantity required to calculate q(Ci,B.) by equation D
is q(Bj). By equation 109, we have
q(B) = XP(Bj) , (F)
where P(Bj) is the minimum of Y(Ci IB) with respect to i.
-05-
- . ' . U .- - .. . .. . .. . . . . . . . . . . . . . . . . . . . . g
P(Bj) = ain (ClB) Y(C 2 IB j ) Y(C 3 1Bj) Y(C4 B
j~i~-l~~3
=~~~ mm C2 1B) D6C 3 1B) 'MT1TB
= mi 1.058 .092 .174 .6761

1(B
I i.058
. ' .085 ' 7.70 ' .687J
= min i , >1 , >1 , .9841
= .984
(G)
p(B2 )
S.143
( .160
.248 .303 .3061
' .140 ' .340 ' .360J
= min .894 >1 , .891 .850)
= .850
143 .248 .3U3 .3061
P(B3 = min
p(3=m .127
.14 ' .236
.2 '
8 .3091.0
,:.3 6.3271
= min > , >1 , .981 .936
= .936
The constant multiplier A is defined by equation 115
6(Bj) I:'
where 6(B) is the conditional probability that a hit is of type

B .,
The determination of q is identical with the procedure of part

VII. From equation 26
A N
we substitute the values of equation A:

5 4 3 2
248q -120q - 47q -22q -16q- =0 (I)
-86-
. *.-
The root is .930 (= qo, say).
The values (Bj) are obtained directly from table 7 by taking the
* ratio of hits of type B. on returning planes to the total number
of hits on returning planes.
6(B I ) S= 294
29 = .737
6(B2 ) = 50 = .125 (J)

55
6(B3 ) = 9 = .138
Substituting the results of equations G, I, and J in equation H,

we obtain:
6(B.)
qo ZO(
930 1.737 + .125 +1381

= . .984 .850 .936
= .930 (1.0433)
- .9703
Substituting in equation F
q(B 1 ) = (.9703) (.984) = .955

q(B 2 ) = (.9703) (.850) = .825 (K)
q(B 3) = (.9703) (.936) = .908
The probabilities q(Ci,B j ) can now be determined from equation D

by using the values given in equations C, E, and K.
6(C.ilB.)
q(CiB) B q(B
i.
-87-
4
. S,*. w. - -~ .. ,. ... -
q(CI,B I ) = (.058) (.955)/.058 = .955
q(C 2 ,Bl) = (.085) (.955)/.092 = .882
q(C 3 ,B1 ) = (.170) (.955)/.174 = .933

q(C 4 ,B I ) = (.687) (.955)/.676 = .971
q(C 1 ,B 2 ) = (.160) (.825)/.143 = .923

q(C 2 ,B 2 ) = (.140) (.825)/.248 = .466
q(C 3 ,B 2 ) = (.340) (.825)/.303 = .926 (L)
q(C 4 1 B 2 ) = (.360) (.825)/.306 = .971
q(CIB 3) = (.127) (.908)/.143 = .806

q(C 2 ,B 3 ) = (.236) (.908)/.248 = .864
q(C 3 ,B3 ) = (.309) (.908)/.303 = .926
q(C 4 ,B 3 ) = (.327) (.908)/.306 = .970
Comments on Results
The vulnerability of a plane to a hit of type Bj on parL C i is
the probability that a plane will be destroyed if it receives a

hit of type B. on part C i. Let P(Ci,B.) represent this vulner-
ability. The numerical value of P(CiB j) is obtained from the
set L and the relationship
P(Ci,B) = 1 - q(Ci,B) (M)
The vulnerability of a plane to a hit to type B. on part C i is

given in table 8.
This analysis of the hypothetical data would lead to the
conclusion that the plane is most vulnerable Lu a hit on the
engine area if the type of bullet is not specified, and is most
vulnerable to a hit by a 20-mm cannon shell if the part hit is
not specified. The greatest probability of being destroyed is
.534, and occurs when a plane is hit by a 20-mm cannon shell
-88-
j
ADOA091 073 CENTER FOR NAVAL ANALYSES ALEXANDRIA VA OPERATIONS EV--ETC FIG 1/3
A REPRINT OF 'A METHOD OF ESTIMATING PLANE VULNERABILITY BASED -- ETC(U)
JUL A0 A BALD
UNCLASSIFIED
ENDlK
CRC-432
"Tl I
on the engine
* 7.9-rn machino
conclusions thi
derived by the
9 guides for loci
prediction of I
tUNRB
Flak, B~
cannon, B2
7.9-Rm
machine
Vulnerability
specified area
a
specified
=101 IM These vulneral

part V, and asi
C is hit, kno,
bThis is the p
hit, When neit:
specified.
MWICOO TESTCHART
RESOLUTO#
NAMPUMMSWEPJ
Of SIANSDARDS.1963-A
on the engine area. The next most vulnerable event is a hit by a
7.9-mm machine gun bullet on the cockpit. These, and other
conclusions that can be made from the table of vulnerabilities
derived by the method of analysis of part VIII, can be used as
guides for locating protective armor and can be used to make a
prediction of the estimated loss of a future mission.
TABLE 8
VULNERABILITY OF A PLANE TO A HIT OF A SPECIFIED TYPE

ON A SPECIFIED PART
Vulner-
ability to
specified
type of
hit when
Forward Fuel area is un-
fuselage Engine system Remainder specified
Flak, B1 .045 .118 .067 .029 .045
20-mm .077 .534 .074 .029 .175

cannon, B 2
7.9-mm .194 .136 .074 .030 .092

machine
gun, B 3
Vulnerability
to hit on
specified area
when type of
hit is un-
.074 .038 .070 b
specified a .114 .179
aThese vulnerabilities are calculated using the method of

part V, and assuming that the y(Ci) the probability that part
C is hit, knowing that the plane is hit, are as follows:
i
v(Ci ) - .084 Y(C2 ) - .128 Y(C3 ) - .212 Y(C4 ) - .576
bThis is the probability that a plane will be destroyed by a

hit, when neither the part hit nor the type of bullet is
specified.
-89-

A Method of Estimating Plane Vulnerability Based On Damage of Survivors" by Abraham Wald (CRC)

Uploaded by

Copyright:

Available Formats

A Method of Estimating Plane Vulnerability Based On Damage of Survivors" by Abraham Wald (CRC)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Method of Estimating Plane Vulnerability Based On Damage of Survivors" by Abraham Wald (CRC)

Uploaded by

Copyright:

Available Formats

What methodology is presented in the document?

What methodology is presented in the document?

What conclusions can be drawn from Table 8?

What conclusions can be drawn from Table 8?

AO-AO9 073 CENTER FOR NAVAL ANALYSES ALEXANDRIA VA OPERATIONS EV--ETC F/6 1/3

REPRINT OF 'A NETHOD OF ESTIMATING PLANE VULNERABILITY BASED -- ETCIU)

NOW .il W, A ..,11

CRC 432 &

A REPRINT OF "A METHOD OF ESTIMATING PLANE

Office of the Chief of Naval Operations COp03) ca iie

16. DISTRIBUTIqN STATEMENT (of this Report)

Approved for public release; distribution unlimited

IS. SUPPLEMENTARY NOTES

aircraft, equations, kill probabilities, probability, vulnerability

This research contribution consists of a series of eight memoranda

DD FORM7 147 EDITION OF.I NOV 65...OBSOLETE

3. Research Contributions are distributed for their

an affiite of the University of Rochester

SNDL Part II:

OpNav: Op-09BH, Op-96, Op-05, Op-50W

Operstions Evaluation Group

CENTER FOR NAVAL ANALYSESt

n TV - Minimu OiC mci u Vine fl the. tAb4Lity

port Vl ampling1. Etot#e,.t....u.4..es4.* 6

Part VIzM' VzZtbflIt$- of A Plane to Differnt typos'

A,, flW4 tt o"~ tno bwe "*a. to accIect~aate cwt

AN EQUATION SATISFIED BY THE PROBABILITfES THAT A

Suppose that pi and Pi (i = 1,2,...) are unknown and our infor-

mation consists only of the following data concerning planes

* The total number N of planes participating in combat.

* For any integer i (i = 0,1,2,...) the number A. of

IThis part of "A Method of Estimating Plane Vulnerability

To simplify the discussion, we shall neglect sampling errors,

From equation 4 it follows that

0 < Pi < 1 (i = 1,2,..., ad inf.). (5)

W;e shall assume that there exists a non-negative integer n such

We shall also assume that there exists a positive inteqer m such

Denote by x i (i = 1,2,...) the ratio of the number of planes

Since the proportion of planes that received at least one hit is

In general, wci obtain

equation 9 can be written

*~ x+ p1 (xi + ... +x i-~=pc i=23..n.(11)

Substituting i - 1 for i, we obtain from equation 11

xi + P~iI(x I + ... + Xi 2 ) - P-c-

Dividing by pl_1 , we obtain

equation 13, we obtain

From equations 11 and 14, we obtain

X.+ Pi x~~A~1 (15)

d.i P.(c. - ci1 ) i (i= 3,4,...,n). (17)

t piqii (i = 3,4r ...,n). (18)

Then equation 16 can be written as

x d1 + t i-i (i =3,4,...,n). (19)

Denote pl(l - a.) by dl, -pa, by d2 and - by t; then we

X: j djtj+jtj+ 2 ti +d i (i = 2,3,...,n). (21)

EQUATION SATISFIED BY ql,...,qn

Ton derive an equation satisfied by ql...qn, we shall express

Substituting i for i - 1 in equation 14, we obtain

obtain from equation 23