This is a reproduction of a library book that was digitized

by Google as part of an ongoing effort to preserve the

information in books and make it universally accessible.

https://books.google.com
 WILS QUARTO
-- 1993 .P761 x 1958
PROGRAM EVALUATION
RESEARCH TASK

PROGRAM
EVALUATION
SEPTEMBER 1958
RESEARCH
TASK •
354

PERT
SUMMARY REPORT
Phase 2

SPECIAL PROJECTS OFFICE

BUREAU OF NAVAL WEAPONS
NOW CALLED : DEPARTMENT OF THE NAVY
PROGRAM EVALUATION WASHINGTON , D.C.
REVIEW TECHNIQUE Dept.
 3

\s E

es
THE LIBRARY
E

MINNESOTA
ΓY OF
TH
OF
SDIANNA

AUS
MA
NU

Wilson Library

For sale by the Superintendent of Documents, U.S. Government Printing Office , Washington 25, D.C. -
Price 50 cents
WILS
1
Quefto
1993
FREIX
1950

TABLE OF CONTENTS

Page

LETTER OF TRANSMITTAL

I. INTRODUCTION TO PERT .
... 1

II . THE INPUTS TO THE PERT SYSTEM 12

III . THE OUTPUTS OF THE PERT SYSTEM 15

IV. FUTURE ACTIVITY 22

APPENDIXES

A. Instruction Manual - Systems and Procedures . ...

. 25

B. Computer Procedures • 45

C. Supporting Mathematics 91

I
 INDEX OF EXHIBITS

I
Page Offic
IMPTOP
A. SYSTEM FLOW PLAN 3 Ballis
Allen
F
B. PERT SYSTEM IN OPERATION 5

C. PROPULSION FLOW CHART . 6 CUSSI

asa
D. INTEGRATED OUTLOOK . 7 Cela
applia
ccasi
E. PROGRESS REPORTING AND TIME INTERVAL ESTIMATING PROCEDURES and it
FOR THE PROGRAM EVALUATION SYSTEM ( PERT) 9
BECCE
F. NORC CALCUL ATOR 12

1.01
G. EVENT IDENTIFICATION FILE 14
tess

H. BALLISTIC SHELL COMPONENT - ALL EVENTS 16 IP

I. PROCE
BALLISTIC SHELL COMPONENT - CRITICAL EVENTS 17

J. ILLUSTRATION OF USE OF THE PERT OUTPUT 18

K. ILLUSTRATION OF RESULT OF HYPOTHETICAL CHANGE 19 1

Cums
El
L. MAJOR EVENT OUTLOOK 20

M. TOTAL REQUIREMENT ( INSTALLATION AND OPERATION ) 24 Sve

tem

stalo
TE
LES

thes

4
sive

12

72

II
 I. INTRODUCTION TO PERT

In January 1958 , the Program Evaluation Research Task ( PERT ) was originated in the Special Projects
Office (SP ) of the Bureau of Ordnance . The project team was directed to the research and development of
improved methods for the planning and control of the complex and far-reaching program developing the Fleet
Ballistic Missile ( FBM ) . The PERT team included members of SP , the contractor's organization , and Booz ,
Allen & Hamilton .
Phase I report , issued in July 1958 , presented the principal theoretical basis of the work and outlined
the proposed installation pattern . While some modifications to the concept have been made since that time ,
the basic concept remains as then discussed . The reader is referred to this earlier report for the fuller dis
cussion of concept .
This report covers Phase II of the project , which has been essentially the development of detail pro
cedures and mechanics , the modification of concept as experience has proved advisable , and the “ prove out "
application of the method in several important areas of the FBM program . This application has demonstrated
considerable value . As a result , the application of the method to the entire FBM program has been authorized
and is proceeding .
With full installation , the PERT method will give a continuous picture of the FBM's status - both all
encompassing and synoptic . SP will have an increased ability to appraise progress and to anticipate poten
tial problem areas while time still remains to take some corrective action .
1. OBJECTIVES OF THE PERT METHODOLOGY . tivities , each tied together in terms of the ultimate
The work described in this report and its prede- system objective . Diagnosis of the status of each
cessor is part of the continuing development of of these activities must be in terms of a common
better management systems by SP . FBil manage- set of rules .
ment has had well -planned reporting and evaluation The PERT system lays out a set of rules for
procedures for some time . PERT extends the ca- such identification . The sole objective of these
pability and builds on the firm , existent foundations rules is to identify those activities in the develop
of the present . ment that inay be potentially troublesome as seen
With this principle in mind , the objectives of at a given point in time. By minimizing attention
the PERT method can be described as follows : on the mass of activities that are of secondary im
( 1 ) The Fustering of increased Order liness and portance at the moment , the managers of the system
Consistency in the Planning and Evaluating of All can concentrate upon those activities which most
Areas in the FBM Project . Order and consistency seriously jeopardize success today .
are basic requisites of a workable management sys- The use of the evaluation system in no way re
tem . To this end , PERT prescribes a comprehen- moves the necessity for wise and forceful technical
sive procedure which is applicable to all areas decision on the part of the FB 11 managers . It merely
within the research and development program . Thus , helps focus their scrutiny . The role of corroborating
the same procedure developed in the identical fash- the suggestion of trouble and the correction thereof
ion is applied to all components within the FBM sys- lies entirely within the province of the cognizant
tem . The method can accommodate activities that technical personnel . The PERT system acts to
are principally of a production nature , as well as conserve their energies for this important work .
those more unpredictable activities inherent in re- A good eva luation system , however , should do
search and development . more than show the current relative importance of
The value of this orderliness and inclusiveness activities . A good system should be capable of
lies in an increased ability to integrate all parts further aiding the managers by showing them some
of the system into a meaningful whole . Evaluations important consequences of decisions that they can
can be made across the boundaries of components make as they seek to prevent or offset potential
and subsystems and thus can achieve validity at trouble .
the system level . The FBM project has uniqueness ( 3 ) The Structuring of a Method To Give Op
in its systems integration . Evaluation must , there- erational Flexibility to the Program by Allow ing for
fore , be aimed at the same level . Orderliness and Experimentation in a Simulated Sense . The conse
consistency of approach enable PZRT to achieve quences of a local decision on more remote events
the same measure of evaluation at both the system are difficult to quickly and accurately visualize .
and component levels . The PERT system provides a vehicle for use by the
However , orderliness , consistency , and inclu- technical man to systematically observe the result
siveness are not virtues in themselves . They are ant of one or more possible decisions on the end
enabling devices upon which an evaluation system objective .
can be erected . Once the technical manager has corroborated
( 2 ) The Providing of an Automatic Mechanism the fact of real trouble within a group of activ
for the Identification of potential Trouble Spots in ities , he is faced with a group of alternative de
All Areas Which Arise as a result of a Failure in cisions . Any one of these options may be capable
One . The PERT system is designed to automati- of correcting an undesirable situation . The eval
cally identify trouble spots and infer the conse- uation system will allow him to correctly appraise
quences of trouble . A complex program such as the the time consequences of these options . Paren
FB :1 system encompasses thousands of separate ac- thetically , it should be noted that the PERT eval

1
uation system will only aid the technical man with page . The events are the numbered circles in Ex
information relative to time . The decision is his hibit A and the activities are the arrow linkages .
to correctly balance the time implications of a Links between events are indicated when there
given action with the related performance and re- is a planned dependency . Thus in Exhibit A , Event
source implications . The PERT system does 51 will not occur until the activities starting with
significantly reduce uncertainty in appraising the Events 52 and 53 are both complete . Event 51
all-important time variable . does not depend on the accomplishment of Event
While the foregoing objectives are noteworthy , 54 , hence no link is drawn connecting these two
they would be of little practical consequence unless nodes .
they could be practically achieved with great speed ( 3 ) Estimation of the Times Necessary To Move
and the maximum of accuracy . from Event to Event Together With A Measure of
( 4 ) The Speedy Hand ling and Analysis of the Uncertainties Involved . A second important feature
Integrated Data , thus Allowing for Expeditious CO- of the PERT system is the gathering of elapsed
rection of Recognized Trvuble Areas . The true time estimates for the activities in the network .
worth of an evaluation system is measured by the This step essentially dimensions the length of the
speed with which it can recognize and correct for arrows for use in subsequent analysis .
out-of-phase situations . In order to fulfill the re
quirement of rapid handling and analysis , the PERT The time estimates used are obtained from the
system has been programmed for the Naval Ordnance technical personnel within the contractor establish
Research Calculator ( NORC ) at the Naval Proving ments directly charged with the accomplishment of
Ground , Dahlgren , Virginia. the activities . The evaluation system is dealing
Analyses and outputs can thus be available to with the future , and the future is uncertain . PERT
the technical managers within a matter of hours from was designed to estimate this uncertainty . The
the time that data is submitted . Answers to some procedure for accomplishing this is probabilistic in
questions may be transmitted via telephone or TWX . form . It depends upon eliciting three estimates of
Thus , there is no need for executive action to be the time necessary to perform each activity from a
delayed awaiting the mechanistic tasks of tabulating , technical man familiar with the activity . These
summarizing , and analyzing of input data . three estimates ( optimistic , most likely , and pessi
mistic ) are transformed to provide the basis for the
2. THE SALIENT FEATURES OF THE PERT statistical inferences relative to the uncertainty in
METHOD . The major part of the theoretical founda- activity accomplishment .
tion of this PERT system was developed early in PERT estimates the future on the basis of in
the study . These principles are described in de- formed judgments . Much can be said about the va
tail in the Phase I report dated July 1958. How- lidity of these judgments . However , the important
ever , it is worthwhile to briefly review the salient fact is that these judgments constitute the most
features of the earlier work in order to appreciate authoritative estimates that exist . Uncertainty will
their impact on the Phase II activity . These basic tend to be high and validity low when events in the
elements are described below . remote future are estimated . Improvements will be
( 1 ) Selection of Specific , Identifiable Events obtained as events come closer to the present . How
That Are Planned To Occur along the way to Suc- ever , it should be noted that if the system should
cessful Conclusion of the Project. A basic feature become manned with some estimators with a low va
of the evaluation system is the choice of a series lidity index , it is able , through an historical study
of detailed accomplishments that must occur in the to calibrate the level of estimators ' validity. No
development program if current plans are followed . system can do more when dealing with the vagaries
These events must be both meaningful to the devel- of an unprecedented future .
opment plan and of a definite, recognizable nature (4 ) Analysis Which Systematically Highlights
and each must be a point in time . Relative Criticality of Events in the Future Devel
It is known that plans will change , and hence opment . In an earlier section , the desirability of
some or all of the events will change during the an automatic flagging of potential trouble spots has
course of time . However , at any point in time , a been noted . The following paragraphs will briefly
definite future plan exists . It is this plan that is describe the procedure for such a highlighting .
typified by the set of events that are specified for The PERT analysis transforms the three time
accomplishment . estimates for the elapsed time to complete a single
(2 ) Linking the planned Events So As To activity into a probability distribution typified by
Graphically portray the Interdependencies among its mean and variance . The procedure then moves
Them . The chosen events comprise a set of ac- systematically through the network of activities
complishments , each of which will exist at a given and events and identifies the mean and variance of
point in time as the culmination of one or more ac- the earliest time for each event , considering all
tivities . The actual work activities carried on in previous events . The mean time is the earliest
the FBM program can thus be visualized as filling expected time at which the event will occur if all
the gaps between events . activities are worked at as soon as they logically
The activities are the links between events . (i.e. , according to plan ) can be . The variance is a
Both activities and events are portrayed in a sys- measure of the uncertainty associated with this
tem flow plan form in Exhibit A , following this earliest expected time of accomplishment .

2
 PLAN
FLOW
SYSTEM

ㅇ
58 55 52

APS
EL ED
te
OF
TIME
51-50
INT ERVAL

EVENTS
CRITICAL
51

3
59 56 53

ㅇ 50

60 57 54
61

A
EXHIBIT
 The final event in the development system (net graphs.
work ) is fixed at a given (desirable ) point in time . ( 1 ) Provision for Initial Inputs tu the System .
The same analysis technique is then applied to the During Phase I , insights into the true nature of the
system in reverse order - taking future events first FBM System were obtained through study of the
and moving back to the present .. This procedure propulsion component of the missile subsystem
evolves a mean and variance of the latest times for ( see Exhibit C, on page 6 ) . Phase II work
each event. These times are latest in the sense involved a more careful reappraisal of the pro
that they represent the latest time each event can pulsion component and an extension of the appli
be accomplished without jeopardizing the point in cation into the flight control and ballistic shell
time chosen as a desirable date for accomplishing components . Input data for evaluation purposes
the final event of the development . were gathered in the field in the course of trips to
Armed with the above knowledge concerning the the Lockheed Missile Systems Division (LMSD ) at
earliest and latest expected times for each event , a Sunnyvale , California .
measure of “ slack " in the plan is determined . Slack These three components were integrated in terms
is defined for each event as being the difference of important common events in the systems , such
between the expected earliest and latest allowable as AX , A1X , BX as shown in Exhibit D , following
times . The most critical events are those which Exhibit C. This procedure enabled the team to ex
have the least slack . Any slip in the planned ac amine this synthes is question on a small scale .
complishment of these events will result in delaying These data were submitted to a hand calculation in
the expected completion date of the final event . The order to draw inferences typical of those that the
relative criticality of all events can be noted by system will produce . Exhibit D is representative of
comparing the magnitudes of their slack values . In the results of the first hand calculations on hypo
addition , when scheduled dates exist for events , thetical data . The black dots represent critical
the evaluation system automatically computes the events within each component and the shaded path
probability of meeting or beating the planned dates. represents the critical path in the integrated out
The reader should bear in mind that throughout look . Thus in this illustrative example , the flight
this discussion a planned rate of resources appli control component would represent the predicted
cation and a specified performance is assumed . critical component in regard to achieving major
Changes in either performance or resources appli event AX . The propulsion component would be
cation rate require reconsideration of the plan and critical in regard to Aix objective and flight con
new time estimates before PERT analysis is made . trols again appear to be critical in objective BX .
The PERT system as conceived during Phase I In addition to the three missile components
is ultimately capable of accomplishing more than previously noted , field work was extended into the
is indicated in the foregoing. However , the princi re-entry body and guidance components . Results of
pal features of the initial installations have been these extensions are now in the process of being
herein described . More complete details will be evaluated and analyzed . In order to test for diverse
found in the explanations shown in the Phase I re aspects of installation , the team also gathered in
port and the appendixes of this report . formation on components outside of the missile sub
Completion of Phase I in this study was followed system . Networks were formed showing the instru
by authorization to effect implementation of the mentation developments required for both the sub
system . Phase II has concerned itself with imple marine and the missile test range at Patrick Air
mentation , implementation techniques , and slight Force Base . These data were gathered in visits to
modifications of the basic model . The following the Interstate Electronics Corporation situated at
section of this report will summarize the activity Anaheim , California .
carried on during Phase II . ( 2 ) Provision for Continuous Inputs for System
Analysis - the Establishment of Reporting Procedures.
3. SUMMARY OF PHASE II ACTIVITY. The Phase The utility of the evaluation system to SP lies in
II work of the PERT team has taken the basic con the stream of outputs that it generates . Continuous
cept as developed earlier, modified it in some inputs are therefore necessary to keep the evalua
aspects , and has taken steps to install the system . tions both current and timely . The PERT team has
It is evident that the best conceived system is recognized that it is necessary to go beyond the
worthless in a practical situation without an effi " one time " data - gathering expedition that validated
cient installation . Consequently , much time has the theoretical approach . A structure of reporting
been spent in carefully formulating installation must be erected for accurate gathering and trans
plans so that all foreseeable impediments can be mission of the input information to SP .
successfully surmounted . A reporting structure has been formu lated and
Exhibit B , following this page , presents a sche detailed by the team . This structure centers on
matic view of information flows and feedback with and is controlled by Sp 12. Field implementation
in the operable PERT system . Phase II work has depends significantly on the efficient activity of
been principally devoted to the efficient achieve -

the SP Field Offices - SPL , SPG , etc. Whereas

ment of the activity implied in the schematic the nerve center of the operation lies within SP
drawing . proper , the sensory extensions lead to the field .
The most important features of the Phase II Official sanction for the implementing of PERT
work are briefly described in the following para lies in an SP instruction issued in mid -October ,

4
 OPERATION
IN
SYSTEM
PERT

M Overall
A
GR
A MGMT
DL
OW
FL
Evaluation
Outlook
INPUTS
Contr
Fromactor
Thread
Critical
OUTPUTS DInfo
.+ etailed
Biweekly
Monthly
Adjustments Technical
Branch

er
NORC Effect
of
change
Computer
Elapsed
Estimates
Time

Proposed
Change

(
11
111
INS
ITUIMILUILIN

EXHIBIT
B
 CHART
FLOW
PROPULSION

151
图
JUNI
第一
度

A
日

9
LA

TTN
LE

图 图 剧 計
11
环

四川
Ho

EXHIBIT
C
 OUTLOOK
INTEGRATED

Control
Fligh t

AIX
BX
Shell
Ballistic

7
Propulsion

TIME
CALENDAR

D
EXHIBIT
(see Exhibit E , following this page ) : This instruc and thus facilitate the installation and administra
tion directs the field offices to issue technical in tion of the evaluation system . This activity has
structions to the various contractors involved . On been pointed in several directions beyond the es
the basis of these " TI's , " contractor activity re tablishment of the basic PERT concept .
ports flow into the field offices and thence to SP
12 .
Reporting forms and reporting frequencies have Intensive indoctrination sessions have been held
been established . The layout of the report form in order to apprise the Sp 12 personnel with the de
consists of a minor revision to the currently opera tails of the approach . The experience gained in in
tive Progress Reporting Form . All information con stallation has been transferred to those who will be
tained on this current form appears on the amended charged with similar work in the future.
one . In addition , provision is made for the three The essential features of the computer analysis
elapsed time estimates - optimistic , most likely , and system control were designed as part of the con
and pessimistic . The form layout was designed to cept . In order to provide for computer liaison,
present the incoming information in a manner that Phase II included a test program for the choice of
was most receptive to the subsequent computer an acceptable individual for special training . This
analysis . individual was given a short course in the NORC
The frequency of reporting was established at operation and tutored in the mathematical bases of
biweekly , monthly, and quarterly intervals . This the system .
frequency will provide the system with its neces Various techincal branches within SP were briefed
sary inputs and , at the same time , will place the on the PERT system . Emphasis was placed on the
least additional load on the contractor . operational aspects of the system during these brief
A manual has been written to carefully describe ings . Procedures for updating , modifying , and ex
the aims and requirements of the evaluation system . perimenting with the FBM development were explained
The manual is designed to facilitate installation in so far as they applied to the PERT system. The
of the PERT procedure. The distribution of the sig nificance of the critical thread analysis was also
manual will include all contractors covered as well examined in detail with representatives of the Tech
as SP personnel and field offices . The manual ap nical Division .
pears as Appendix A to this report . The future personnel and computer requirements
(3 ) Selection of a Computer Facility and Program for SP were studied and portrayed . These require
FotTiulation . The importance of speed in analysis ments were based on a rapid extension of the PERT
has already been emphasized . Phase II activity in system throughout the entire FBM development . The
cluded the task of selecting a computer facility as estimated personnel requirements were spelled out
well as the programing of the computer to accommo as were the costs occasioned by future computer
date the PCRT analysis . usage . Chapter IV develops a detailed installation
A survey of Navy installations within the Wash plan and personnel requirements to make a com
ington environs resulted in an SP transfer of funds plete installation of PERT in a six -month period .
to the Naval Proving Ground at Dahlgren , Virginia , The planning described above was done in the
for machine time on the Naval Ordnance Research face of incomplete information owing to the fact
Calculator, the NORC. The NORC was most that the system has not been widely installed . Ef
readily available and appeared to provide the ficient operation of the evaluation system will re
greatest economies of calculation and analysis quire that planning be continuous . Thus , the in
among the available computing establishments. A itial activity undertaken by the PERT team must
picture of the NORC appears in Exhibit F , on be constantly reviewed and updated in the light of
page 12. experience .
A computer specialist on the PERT team care (5 ) Continuance of the mathematical Appraisal
fully monitored the programing activity carried on of the model . The evaluation system has aimed at
at Dahlgren . A successful and efficient program striking a balance between strict mathematical va
has been formulated . The NORC capability was lidity and operational facility . Operational con
demonstrated to SP managers in early October siderations dictated certain features of the system
when actual analyses were run on valid FBM data in spite of the introduction of analytical biases .
gathered during the Phase II effort . Phase II has continued the original mathematical
A most important feature of the computer pro work to evaluate the importance of the recognized
gram as developed lies in the fact that it is com bias. Results of this work will be issu ed as a sup
pletely general in nature . This feature permits plemental report .
FBM system changes to be made very simply and 車
without requiring computer reprograming . Further ,
it can be used in the analysis of any R&D pro
gram, provided the inputs are in the form as devel This chapter has given a synopsis of the range
oped by PERT . of activities undertaken by the PERT team during
( 4 ) Planning for SP Management. The efficient both Phases I and II . The succeeding chapters of
installation of a significantly new system demands this report will describe in greater detail the work
careful planning . During Phase II , the PERT team involved in Phase II . Technical details are pre
has worked to accomplish the necessary planning sented in the appendixes .

8
 DEPARTMENT
NAVY
THE
OF SPINST
7720.1 OFFICE
PROJECTS
SPECIAL
Office
Projects
Special INST
7 720.1
.SP 1958
October
16
,D.
25
C.
Washington W12
Sp
:a F
- em
October
16
1958 Level
Contractor
.at
System
Reporting
PERT
the
of
Installation
.
4

.
a
Requirements
Basic
and
Plan
7720.1
INSTRUCTION
SP
be
will
system
evaluation
program
this
of
The
installation
S,From
:Dpecial
irector
Projects .tof
consist
will
Team
The
approach
ofeam
use
aby
accomplished
D istribution
:To the
,a nd
office
field
Sp
appropriate
the
from
representatives
office
appropriate
other
analysis
,orrogram
p
planning
contractor's
Procedures
:Progress
Subj
Estimating
Interval
Time
and
Reporting
for designate
).may
(as
contractor
the
P
( ERT
System
Evaluation
Program
)the
designated
the
between
divided
be
will
Sp
within
Responsibility
Time
:(1)Report
Encl
and
Progress
of
Estimates
Interval assigned
.as
branch
technical
mcognizant
the
of
and
aSp
12ember
analyst
Introduction
for
basis
the
forms
which
System
Evaluation
Program
The
,1. system
are
:Basic
the
of
requirements
instruction
this
PERT
and
conceived
was
result
PERT
the
of
.is
project
Office
Projects
Special
the
of
management
provide
to
developed
)w(S ith
P which n
accomplishmen
efinable ts

6
devents
(list
s)oof
1Aelect
icontinuous
appraisals
, ntegrated
of
progress
program
the
and
date
to reported
;be
will
progress
components
FBM
various
for
outlook
,singly
.progress
combination
in
and
system
eprovides
re
for valuating
asis
appropriate
-abThe
program
plans between
interconnection
the
outlines
which
chart
flow
(2)A s
future
accomplish
to
in
objectives
major
FBM
of
light
the
Program's ;
events
commodity
important
most
time
,
activity
the
perform
to
required
time
of
(3) stimates
E
progress
.T
Purpose
2.
forth
set
o
requirements
reporting
procedures
and a
diagram
;flow
the
in
events
between
connections
bynd
represented
responsibilities
,and
PERT
identify
information
gathering
for
from
contractor
andthe
instruction
,tIdata
isnitially
his
.reporting
Sp
to estimates
time
interval
and
progress
of
reports
(4)Periodic
INSORD
to
applicable
the
and
Sunnyvale
contractors
subsystem
missile .(1) nclosure
E
within
their
cognizance
extended
As
is
system
PERT
the
.to
other
,the
.contractors
apply
will
herein
contained
procedures .
b Events
of
Selection

instruction
manual
developed
experience
from
inAn
gained
early accomplish
definable
represent
must
system
this
for
selected
Events
methodology
PERT
of
applications
be
will
system
missile
the
to be
c the
of
Definitions
established
.date
can
ompletion
which
afor
ments
.
personnel
cognizant
to
provided aFBM
.F
Spby
required
events
revised
)new
or
ithe or
re
( ncluding
he
action
tavailable
,are
plans
management
program
where
components
E
(1)h
beennclosure
Formatas
developed
Reporting
.3. the
as
acceptable p roduction
.g.
records
company
various
,(ewith
together
milestones
interval
time
and
data
progress
of
reporting
contractor
initiating
for
means selec
afor
sources
as
used
tbe
plans
e ill
ngineering
)w, entative
schedules
PERT
for
estimates
.n
Ithose
is
4153
Form
NAVEXOS
where
cases
.by
team
the
events
PERT
of
tion
,e
used
(currently
an1)t nclosure
emporarily
represents
reporting
additional
surpose
ulti
mrequirement
is
,a-p
and
form
indicated
.Icolumns
ctertain C.
Chart
Flow
the
Establishing
need
or
contractor
the
by
out
filled
be
not
.
office
field
FBM
Program
the
of
component
each
for
is
chart
required
flow
One
installation
.of
time
the
at
Team
by
designated
as

E X H I BI
E T
 7720.1
SPINST SPINST
7720.1 OFFICE
PROJECTS
SPECIAL
OFFICE
PROJEC
SPECIALTS
1958
October
16 1958
October
16

individuals
technical
more
or
one
of
assistance
the
,with
team
The
develop
the
in
engaged
actually
group
technical
the
of c.Supplemental
Reporting
head
by
assigned
c
a
events
of
(2),list
final
d
wwork
mentthe onnect
nd
etermine
(1):, ill
events
of designated
certain
on
Additional
requested
be
may
information
network
T his
relationship
.planned
their
indicate
to
events
the personnel
Sp
cognizant
by
necessary
deemed
as
chart
flow
the
of
areas
.chart
flow
the
is
established
so for
used be
possible
(1)w,E
offices
.as
field
and ill
nclosure
nsofar
ISp
.may
be
chart
flow
cases
,t
entire
datahe
certain
In
the
reporting
.
d Estimat
Interval
Time es
Obtainin g
by
Sp
the
approved
form
when
transmittal
as
used
and
reproduced
the
on
System
Evaluation
of
installation
the
in
step .
office
field
next
The
che
for
intervals
time
of
estimation
the
involves
base
contractor
the
on
events
defined
between
connections
the
by
represented
activities
Stratan
.
chart
flow

W. F.
RABORN
.
5 Reporti
Recurre res
.Procedung
nt

.
a Responsibility
:
Distribution
w
bieekly
-the
of
preparation
for
responsible Spz
is
contractor
The
to
use
,ifor
its
instruction
w sith Sp3
E (1)The
form
. nclosure
report
completion
for
scheduled
or
completed
events
all
of
status
the
contain
events
future
any
of
status
and
perio
w eek
)(2two
theprevious
during
anticipated
.is
data
completion
earlier
or
adelay
where

01
.
b Frequency
report
wepkly
period he
i
t:h,wis
bi
reporting
established
The
1:F
on
E of
copies
.weeks
alternate
Tuesday
d our
ue
(1) )nclosure
12
,to
co
as
beSp ne
follows
2)two
are ( opies
submitted
report
cach
cog.z
:?to
c
o
Branch
Technical
(1),Sp
cognizantne
opy
c::opy
Office
.Field
Projects
Special
require
will
chart
flow
the
to
event
an
of
revision
or
addition
The
estimates
interval
time
additional
and
dates
completion
new
of
reporting
on
provide
to
is
contractor
.estimates
existing
of
revision
Trhe
/o
and
activity
an
for
re
-or
eaddicions
known
wall
report
eekly
- stinates
bi
the
r
-as
to ange
short
well
future
applies
This
.activities
of
group
or
after
due
report
-w
bi
the
in eekly
first
appear
are
data
These
.
toevents
known
,become
they

E
EXHIBIT
 (PERT ) CLASSIFICA :
REPORT PROGRESS & TIVE INTERVAL
ESTIMATES
From : (Name & Location of Contractor ) To : Contract No. Report Period
From /To :
Flow Chart No.

For office use Activity Identification Time Int . Est. Do

only ( Do not Proceding Succeeding Opt Most Pess Couple Not
rui in ) Event No. Evont No. (wks) Lkly (wks) tion F111 REMARKS
(wks) date

G
)
( 1 ) ( 2) (3) (B) (c) (D 1 (E ) (6 ) (6 ( )

Signature and Mtlo of Responsible official: Dato Signed :

• Columns D, E, & F : These ostimates should CLASSIFICATION :

be given for the full activity even though
the activity has already startod.

INSTRUCTIONS FOR COMPLE : ' OR OF LE PORT OF

PROGRESS AND TIE INTE AL ESTDATES TO'M
GENERAL: This report is designed to provide information COLON E : Entor he not likely or most probable time intor
on program progress and estimates of time intervals val for an activity in terms of weeks ( to the nearest tenth.
required for events included in the PERT flow chart . of a week ) defined as : ( 1 ) . tine longer than or equal to
A separate report will be submitted for each PERT flow that entered in Column D ; ( 2 ) the time that would be reason
chart . Tine interval estimates should be made by the ably be expected by the person best qualified to judeo; or
person in the highest level of the technical organization ( 3 ) the time that would occur most often if the activity
who could perform the work hinself. The best judgment could be repeated numerous times under exactly the same
of those responsible for the work is desired. conditions ,

Copies will normally be submitted in writing to be re- COLUMN F: Enter the pessimistic time interval for an acti
ceived in the Sp office by the close of each alternate vity in teras of weeks ( to the nearest tenth of a week )
Tuesday . Required timing will be specified at the defined as : ( 1 ) Tire required i significantly worse luck
installation of the Program Evaluation System for the than usual is encountered ; ( 2 ) a time not influenced by
particular component involved . Reports are required strikos or by catastrophes or " acts of God , " such as foot
even though no events were completed or scheduled for fires , etc ; or ( 3 ) a time which no reasonably qualified
completion during the previous two ( 2 ) week period . person or porsc .3 would expect to be exceeded except in
Verbal reports will be accepted under unusual circum- very unusual circumstances. Divided opinion among experts
stances and must be followed up with a written report . as to the success or failure of a project makes it neces
Four (4 ) copies should be forwarded as follows : sary to include the effect of a fresh start upon the pessi
estic time estimate .
1. Special Projects Office , Bureau of Ordnance
Navy Department, Washington 25, D. C. Attn : Sp12 (2 ) COLUMN G: Report the dato of completion ( actual or schedul.
Sp Cognizant Technical Branch ( 1 ) ed ) for al activities completed or scheduled for comple
tion since the last bi - weekly report ,
2. Cognizant Sp field offico . ( 1 )
COLUMN H : Explain the reason for changes made in schedule
COLUMNS 1 , 2 , 3 , & 4: Do not fill in . dates and time interval estimatos . Whenover an ovent or
dlestone is ollainated , added , or description is changed
COLUMNS B & C : Enter the now chart identification from the flow chart designation , such modification should
numbers for the events which form the beginning and be reported along with complete new information for other
onding points of the intervening activity currently events and activities which may be affected by the change-
being reported . The proceding ovent lies to the left 1.6. , in torns of completion dato or time interval estimates
or tail of the arrow on the PERT flow chart . together with the reason for change , using a supplenental
shoot necessary .
COLUMN D : Entor the optimistic time interval for an
activity in terms of vooks ( to the nearest tenth of a
vook ) dofined as : (1 ) An attainable time when sheni
ficantly bottor luck than usual 1s expectad ; or ( 2 ) »
dnima tine with practically no hope of completing the
activity in loss than optidstic tio.

EXHIBIT E

11
 NORC CALCULATOR

EXHIBIT F

II . THE INPUTS TO THE PERT SYSTEM

The bare concept exists, has been tested, and tem and should provide a basis for the designation
has proved useful. Two data inputs must be carried of individuals from both contractor and Insord as
out to get the complete system under way . First is members of the team for the given component in
the initial data requirement to actually complete the stallation .
installation across the entire FBM project. Second Additional details should then be covered with
is the provision of continuing, updating data once the contractor and Insord team members. The
the installation is complete . SPINST 7720,1 ( Exhibit E) and the instruction man
The experience of the PERT team during Phase ual ( Appendix A) should be carefully studied so
II furnishes guidelines for accomplishing these that the detail implications of the system are under
inputs. stood .
1. INITIAL INSTALLATION OF THE EVALUA ( Included in Appendix A is a glossary of terms
TION SYSTEM . The steps to be taken in an initial used in the PERT approach . The reader may wish
installation in a component area are as follows: to refer to this in reading the ensuing material. )
( 1) Briefing of Contractor and Insord Repre ( 2) Preliminary Event Sequencing. The SP
sentatives. The contractors possess the most inti- team members ( an evaluation man and a technical
mate knowledge of the current state of their devel- man ) will have previously constructed a rough flow
opment activities . The cognizant Insord office has chart of the given development process. This flow
the responsibility for the field monitoring of these chart should now be subjected to the scrutiny of
activities. Therefore, as the first step in an initial the remainder of the team . Modifications will be
installation, a PERT team should undertake to indoc- made in the light of knowledge of the latest de
trinate responsible representatives of both in the tailed plans .
evaluation system requirements . These briefings The full team should then carefully connect the
should assume a general understanding of the sys- events so as to constitute the network . The connec

12
tions between events are evidence ofdependencies or concerned. Such a review represents a necessary
interactions . These dependencies can be of two kinds . precaution for the initial incoming information from
They can be direct technological dependencies ( where a contractor. A typical flow chart appears in Ex
one event cannot physically occur until some previous hibit C.
activity has been completed . ) Or , the dependency may
.

simply be a planning dependency dicated by ef 2. UPDATING DATA FOLLOWING SYSTEM IN

ficient use of time . Either condition warrants the STALL ATION . The dynamics of the FBM develop
drawing of an “ activity arrow . " ment require a continuous updating of flow chart in
( 3 ) Time Estination and General Authentication . formation after it has once been obtained . The
To this point , the contractor technical personnel PERT system provides for this modification by pre
(other than possible team members) have not been scribing periodic reports from the contractors . These
approached . This avoidance of contacting techni reports are transmitted from the Insord office to the
cal personnel has been deliberate , since the team control center in Sp 12. The actual updating takes
should minimize the amount of time in which the the following form .
technical people are diverted from their primary job. ( 1) Biweekly Updating. At biweekly intervals
However, once the team has sequenced the network the contractor is required to report on a standard
to the best of their own knowledge , authentication form ( see Exhibit E ) the status of all events that
must be obtained from the most knowledgeable have been completed or scheduled for completion
person - the technical man charged with the de during the past two weeks . In addition , the con
velopment. tractor can report on any other events when their
After review and approval of the flow chart, the anticipated time for accomplishment changes .
technical specialist must provide the three time As these modifications are posted within the
estimates required for each activity link in the net PERT accounting , SP will be continuously aware
work . Careful explanations should be made in re of the position of all phases of the development.
gard to the significance of these optimistic , most The biweekly updating concerns a relatively
likely, and pessimistic times. Attention should be few events that are scheduled for current completion
specially focused on the obtaining of best current However, this information will generally provide
estimates of elapsed times between pairs of events. little basis for making inferences about the course
In some cases , the specialist may indicate the
of future development. Consequently, a monthly re
existence of optional developments in his planning. quest will arise from SP asking for more detailed in
One of several options will be adopted depending formation .
upon future research . This situation can be ac ( 2 ) monthly l'pdating . The monthly request will
commodated in the evaluation . The team members ask for re - estimates of certain activity times in the
should obtain time estimates for all options and network . Re -estimates will be required for all ac
note the currently preferred one if such is the
tivities that lie along the most critical path as de
case . termined by the evaluation analysis . In addition ,
Once a flow chart has been developed and has certain other activity times may be requested by
received authentication and time estimates have SP in order to determine the relative validity of
been rendered , the team must carefully document times previously submitted by the contractor. This
the identification of each event. A file should be latter type of request can be used as the sample
set up to record these identifications. They should control audit of SP .
be complete and precise so that subsequent review All requests for the monthly updating must be
will not be faced with a set of vague meanings. transmitted by Sp 12 , although they may originate
The event identification file will be preserved in any SP office. The breaching of this requirement
both at SP , Insord , and within the contractor's es will only result in placing an unduly heavy load on
tablishment . The file can be very simple , as long the contractor and confusing the evaluation system .
as it presents the necessary information ( see Ix The request for the me hly updating is made
hibit G , following this page ) .
.
the standard reporting for m ( see Exhibit E ) . The
(4 ) Fields of Initial Application . The foregoing identical form , when properly completed by the con
sequence of steps has been followed in the first
tractor , will comprise the required report .
installations of the evaluation system . Installa
The monthly updating provides SP with the ve
tions have been completed in several component de hicle for examining at frequent intervals those ac
velopment areas of the missile subsystem and also
in certain nonairborne instrumentation development tivities that are currently most critical in the
areas . The first installations have been concen achievement of the end objective . At the same
trated on the missile sub system in recognition of time , it does not require an abundance of informa
tion about activities whose status is of relatively
the current importance of this subsystem . Thus, little concern. Eoth the contractor's reporting
initial flow charts have been completed for pro
load and SP's analysis load is thus minimized with
pulsion, flight control , ballistic shell , and re- entry
1

body . Work is currently progressing toward the out sacrificing significant information .
inclusion of the remainder of the missile . ( 3 ) Quarterly Updating . The monthly updating
The flow charts and time estimates for these requests will not deal with the majority of the
components have been subjected to an independent events in the FBM system . However , all events
reasonability review by the SP Technical Branches must be reached before the development is com
13
plete. Activities differ in importance in only a tion procedures have been noted. The manual
relative sense. It is , therefore , necessary to peri- ( Appendix A to this report) develops the system
odically evaluate the totality of the network . This atic step -by -step procedures that must be provided
is accomplished in a quarterly updating report. for during the installing and updating phases.
The quarterly updating is essentially a recon 3. CONTROL AND STABILIZATION OF INPUTS .
struction of the original flow chart complete with The PERT method will deal with masses of infor
re- estimates of activity times . The quarterly up mation emanating from different sources and arriving
dating should generally follow the procedure set up at different times . Therefore, it is of utmost im
for generating the original flow chart . It will be ac
portance that procedural checks be set and strictly
complished with SP participation along with the followed .
contractor and Insord . Some briefings that were
necessary in the original installation will be found In order to assure that all information is received ,
correctly handled , and transmitted, a set of logs ,
to be superfluous in this updating . editings , and checks and balances has been de
The results of the quarterly updating will provide signed . These are outlined in Appendix B , “Com
SP with information to re -evaluate the entire devel puter Procedures . " Adherence to this design should
opment and thus obtain a fresh prognosis . minimize the opportunity for trouble arising within
As the PERT method is applied to more and more the SP system . Most importantly , however , should
areas of the FBM system , it will be necessary to trouble arise , the system itself will give positive
carefully plan the times at which reports will be indication of such trouble . Subsequent corrective
submitted . The report cycles should be staggered activity can then be directed toward a precise area
for the various subsystems and components . The rather than a search over all possible sources of
biweekly reports should occur on different weeks trouble .
for different components. The monthly and quar The set of controls must encompass transmis
terly updatings should similarly be scheduled at sions to and from the computer . The details of
different times . Such scheduling of reports should these controls are presented in Appendix B of this
make for a constant work load with little week; to- report, which deals with the elements of the com
week variation . puter installation and analysis .
This duscussion has avoided many details that With the dat a-gathering structure developed , it
are necessary for a workable installation of the sys- is appropriate to discuss the output side of the
tem . However, the main ingredients of the installa- PERT system .

EVENT IDENTIFICATION FILE

SUB SY ST EM 27 - MISSILE
-

COMPONENT 400 - BALLISTIC SHELL

EVENT DESCRIPTION

27-400-024 Compl . Modif, and c/o by Grd. Tests Bal . Shell

( AIX )

27-400-025
Arrival of Vehicle at PAFB ( AIX )

27-400-026
Flt . Test Bal . Shell (AIX )
27-400-027 Comp

27-400-028

EXHIBIT G

14
 III . THE OUTPUTS OF THE PERT SYSTEM

The output side of the evaluation system is the computed in the run and permanently kept , although
payoff of the system. Understanding and profitable not currently printed. These suppressed values can
utilization of output is a pre-requisite to the matur- be printed with very slight modifications within the
ing of the overall procedure . printing run.
1. OUTPUT POSSIBILITIES . The PERT system The computing run calculates the variances of
the earliest and latest time distributions for each
is capable of presenting a large variety of informa event . These are not shown on the output , but are
tion in many different forms . An initial computer used in probability analyses in relation with col
output format has been designed. It is important to umn 6 , headed Pr .
note that this form does not present the full range At certain times when optional plans exist , it is
of information that can be printed if desired . How
e ver , it does exhibit those factors that will be most important to typify the resource and performance
levels projected for the different plans ( see Appen
significant to the SP manager at this time . Subse di x A ) . When such levels have been indicated for
quently , as experience and understanding mounts , the activity sequences , their values are recorded
the format may easily be altered to enable a more on the output cards and are available for examina
complete exploitation of the methodology . tion, if desired . In addition , certain basic proba
( 1 ) Printed Outputs . The form of the printed bility calculations are made in the computing run
computer output appears in Exhibit H , following this and added to the historical record . Although these
page . All printed data is numeric as the NORC com calculations are not currently being used , a pro
puter ( which produces the outputs ) does not work ef
cedure does exist for their utilization in having the
ficiently with alphabetic characters . computer designate a best set of scheduled dates
The first column on the output sheet shows an ( see Appendix B ) .
event code that has previously been assigned in the ( 3 ) Output Options . Certain outputs have been
process of flow charting. This is a unique designator. prescribed and are currently being printed on ma
Systems , subsystems, and components can be dis- chine runs , as indic ted in Exhibit H. Other y alu es
tinguished by their code numbers . Translation of are recorded internally within the machine memory
the event codes can be easily accomplished with a and can be presented with some little provision for
listing from the previously noted “ Event Identifica- the change . There is a third possibility that should
tion File " ( see Exhibit G. ) be described . It exists as an option that can be
The second and third columns on the output re exercised with the proper setting of switches on
port form show the values of Te and Tl . The the machine.
entri es in these columns are the result of machine The totality of all events can be presented as
computation . They indicate the expected values the regular output of a machine run . This detailed
( means ) of the earliest and latest time distributions information will be of interest to personnel handling
for each of the coded events . Roughly speaking , the various components . Exhibit H shows a ty pical
the Te is the earliest time at which the event will output of a so - called "regular " run .
occur in terms of current expectations . The value For some purposes , the mass of detail shown in
Tl is the latest time the event can occur without the " regular " run will be excessive for quick apprais
disturbing the date of the system objective . Printed al . Therefore , it is necessary to adopt some crite
values for Te, TL, and Ts are measured in weeks rion for extracting from the " regular " run those
from the date printed at the top of the forms . particular aspects that are significant . The machine
The fourth column shows the difference between has been programed to present only the most criti
the values in columns 2 and 3. This column (de cal events for the analyst's perusal . With the sim
scribed as slack ) will be non- negative in the pres- ple throwing of a switch , the output edit run will
ent computations which set the latest time for the suppress the data on all events with large amounts
system objecitve at the computed earliest time it of slack , Exhibit I , on page 17 , shows how the
will occur. If, in determining a TL , the system previous data are now presented with this discrim
objective ( end event) is set at a time beforeits ear inating switch on
liest expectation , Te , then column 4 can show neg The upper portion of Exhibit I merely shows
ative values .
The fifth column indicates the scheduled time for those events within the ballistic shell component
an event, if such exists . It appears in the output that have zero slack . These events are the most
as a coded value in the same terms as Te and TL . critical to the development of that component. Con
( The last column on the sheet showsthe same value sequently they are of the most current interest .
in decoded form , i.e. , as an actual date . ) Another setting will enable the machine to print the
The sixth column describes the probability of output information on all events that have a small
meeting or beating the scheduled date in terms of amo unt of slack . The lower portion of Exhibit I is
current plans and anticipations. typical of such a print- out . Here , all events that
( 2 ) Suppressed Values . A record is kept in the have slack within the range zero to five weeks have
computer of all transactions that are submitted for been laid out for special study . Thus , this optional
analysis . In addition, other pertinent factors are study will give the technical manager data relative

15
 PERT SYSTEM BALLISTIC SHELL COMPONENT ALL EVENTS

71258 0.0 1
DATE WEEK PAGE

OOOONO
EVENT TET, Ti -T Ts P. Schodulod Dato

27-400-000 0.0 0.0

27-400-001 3.2 3.2
27-400-002 3.2 3.2
27-400-00 ? 7.5 7.5
27-400-004 4.2 15.5 11.3
27-400-005 0.0 0.2 0.2
27-400-006 3.0 3.0 0.0
27-400-007 3.0 3.2 0.2
27-400-008 3.0 3.2 0.2
27-400-009 3.5 3.5 0.0
27-400-010 12.0 15.5 3.5
27-400-011 7.0 21.5 14.5
27-400-012 7.5 7.5 0.0 9.0 .99 9-12 - XX
27-400-013 15.5 15.5 0.0
27-400-014 10.0 30.7 20.7
27-400-015 26.0 46.7 20.7
27-400-016 35.5 35.5 0.0
27-400-017 21.5 21.5 0.0
27-400-018 15.5 110.3 94.8
27-400-019 15.5 110.3 94.8
27-400-020 30.0 50.7 20.7
27-400-021 51.5 51.5 0.0
27-400-022 30.0 50.7 20.7
27-400-023 33.0 53.7 20.7
27-400-024 54.5 54.5 0.0
27-400-025 55.0 55.0 0.0
27-400-026 58.0 58.0 0.0 50.0 +0.00 6-30 - xx
27-400-028 70.3 116.3 46.0
27-400-029 96.3 96.3 0.0
27-400-030 40.0 89.3 49.3
27-400-031 56.0 105.3 49.3
27-400-032 60,0 109.3 49.3
27-400-033 60.0 109.3 49.3
27-400-034 || 112.3 112.3 0.0
27-400-035 | 112.3 112.3 0.0
27-400-036 82.3 82.3 0.0
27-400-037 63.0 112.3 49.3
27-400-038 || 116.3 116.3 0.0 128.00 .99 12-31 . XX

EXHIBIT H

to those activities and events that are in the second tem managers in devising modifications to the plan .
most critical range . There are many ways in which the PERT out
In summary , only the immediately pertinent infor puts can be used by those charged with the success
mation appears on the current output sheets . How ful development of the FBM . It is not possible to
ever, the range of output information can be ex itemize all the ways in which the outputs can be
panded significantly as experience allows for its profitably utilized . However, some of the principal
profitable utilization . types of inferences can be noted .
This section has described the values that ap ( 1 ) Determining Relative Criticality of Events .
pear on the output form , as well as those that can The "regular " output of the NORC lists the slack
appear. The following paragraphs will describe the time currently associated with events in the devel
type of inferences that can be drawn from the out opment . Thos e events with the smallest slack are
put sheets . the most critical events . Slippages of these events
will cause the expectation of the final event to slip .
2. INFERENCES FROM OUTPUTS . The FBM de Consequently , the managers should carefully moni
velopment proceeds in accordance with a plan . The tor progress in these areas .
PERT evaluation appraises the validity of that plan Contrariwise, events exhibiting large amounts of
in terms of its timely accomplishment . The evalua slack are in relatively good shape. The evaluation
tion procedure also can give guidance to the sys is a relative one. Surveillance over all activity is

16
necessary, but, the relative importance of activities ance specified . The PERT runs can give the tech
as shown by the slack computation enables the nical personnel insights into the varied consequences
cognizant branch to logically allocate its super- of these available options as they affect time . (of
visory activity . course knowledge as to the possibilities of resources
( 2) Appraisal of Schedule Feasibility . The or performance trade-offs must come from the tech
existence of a definitive plan includes the designation nical personnel involved or other sources external
of specified times for the accomplishment of events - to PERT. )
that is , a schedule . These scheduled times are nec- Exhibit J , following this page, illustrates the
essarily set far in advance of actual accomplishment. use of the PERT output for this purpose . An op
In some cases , it is desirable to change a scheduled tion might exist for shortening an activity time by
date to benefit the overall program . This is possible five weeks . See event No. 27-400-013 . The effect
only if the potential slippage can be seen at an of this shortening on an objective event is sought.
early enough date. The PERT experimental output indicates the results
The PERT system provides the managers with a that would be obtained by this change in the large
probability measure that describes current chances typescript. As shown, the proposed improvement
of meeting future scheduled dates. A low probabil- has no effect in reducing the time in achieving the
ity of meeting a scheduled date ordinarily will gen objective, since it remains at 112.0 weeks. Hence ,
erate remedial actions which will improve the ob- it would not be wise to expend additional resources
served probability. However , if thorough an alysis in the given area as they would be unproductive in
indicates that no effective remedy is available, gaining a time saving . It is possible to draw a gen
knowledge of the existence of the low probability eral conclusion , or “ principle , " from study of this
can be used to alter the schedule so as to more ef- and similar cases . It becomes evident that time im
ficiently use the resources available to the program . provements within slack areas produce no effect on
( 3 ) Experimentation . The existence of criticality the timely accomplishment of the end objective.
or potential slippage is a flag that causes the sys- Exhibit K , following Exhibit J , illustrates the
tem to look to available remedies for an undesirable result of another hypothetical change. In this case,
situation . These remedies can be alternate devel- the time improvement was made in an activity that
opments , changes in the rate of resources applica- was of a critical nature , as noted by the zero slack
tion , or variations in the level of technical perform- ( event 17-400-024 ). The result of this reduction is

PERT SYSTEM BALLISTIC SHELL COMPONENT O

CRITICAL EVENTS

DATE 7 | 12 | 58 WEEK 0.0 PAGE .

EVENT TE TT-TTS २. Scheduled Date

27-400-006 3.0 3.0 0.0

27-400-009 3.5 3.5 0.0
27-400-012 7.5 7.5 0.0
27-400-013 15.5 15.5 0.0
27-400-017 21.5 21.5 0.0
27-400-016 35.5 35.5 0.0
27-400-021 51.5 51.5 0.0
27-400-024 54.5 54.5 0.0
27-400-025 55.0 55.0 0.0
27-400-026 58.0 58.0 0.0
27-400-036 82.3 82.3 0.0
27-400-029 96.3 96.0 0.0
27-400-034 112.3 112.3 0.0
27-400-035 112.3 112.3 0.0
27-400-038 116.3 116.3 0.0

27-400-001 0.0 3.2 3.2

27-400-002 0.0 3.2 3.2
27-400-005 0.0 0.2 0.2
27-400-007 3.0 3.2 0.2
27-400-008 3.0 3.2 0.2
27-400-010 12.0 15.5 3.5

EXHIBITI

17
 PERT SYSTEM
DATE 7 | 12:58 WEEK 0.0 PAGE

EVENT TE T TiTTS
- P, Scheduled Dole

27-400-013 15 21.5 12.0

10 27.5 17.0 Activity reduced 5 weeks in duration

27-400-38 112.0 112.0 0.0

112.0 112.00.0 No effect on objective
PRINCIPLE : TIME IMPROVEMENT IN SLACK AREA
PRODUCES NO EFFECT ON OBJECT

EXHIBIT J

in fact beneficial. Again , a principle can be ad On the contrary , wnen slack exists it may be pos
duced . An improvement along a critical path can sible to strive for superior performance or to extract
produce an effect on the end objective up to the resources for a more urgent application .
amount of the original improvement . When the PERT system is used as part of such
experimentation , it essentially “ simulates ” the fu
The foregoing principles can be summarized in a ture development process . The NORC speedily re
slightly different fashion . Resource and perform veals the results of the simulation . The cognizant
ance trade-offs will tend to be productive when ap personnel can then make decisions with a more
plied to activities that lie along the most critical complete awareness of the consequences of those
path . When slack exists in a sequence of activ decisions . These consequences show not only the
ities , nothing is to be gained by degrading perform local effects but also the general effects on the
ance or increasing the rate of resource application . integrated FBM development .

18
3. SUMMARY PRESENTATIONS . Top management this flexibility. Exhibit L , following this page,
must have a means for quickly grasping the essen demonstrates a presentation that has been developed
tial status of the total program . Whereas the in for showing the outlook for major event accomplish
dividual technical branches are concerned with ment.
parts of the whole , top management must have pro In Exhibit L , the wealth of detail has been sup
cedures for synthesizing the entire system . The pressed and only events of pre- eminent importance
PERT method provides procedures for this integra are shown . ( The data shown in this exhibit are ar
tion . tificial . ) In this exhibit the current outlook is vis
( 1 ) Major Event Outlook. The NORC program is ually compared with both the schedule and the last
capable of extracting different levels of detail from previous information . Thus, current progress can
the analysis . Previous sections have described be quickly compared with desired progress . In ad

PERT SYSTEM
DATE z 12:58 WEEK 0.0 PAGE

EVENT TE ,T T.-T. Ts P. Scheduled Dole

27-400-024 35.0 35.0 0.0

30.0 32.0 2.0 Actifity reduced 5 weeks

27-400-038 112 112 0.0

109 109 .0 Effect on end -objective 3 weeks saved

PRINCIPAL : An improvement along the

critical path can produce
up to the amount of improvement

EXHIBIT K

19
 MAJOR
EVENT
OUTLOOK

AIX

SCHEDULE
)(Date

PERT
FROM
ESTIMATE
CURRENT

(Date
PERT
FROM
ESTIMATE
PREVIOUS
)LAST

BX

20
DSCHEDULE
)( ate

PERT
FROM
ESTIMATE
CURRENT

(Date
PERT
FROM
ESTIMATE
PREVIOUS
)LAST

DATES
CALENDAR

L
EXHIBIT
dition, the current position is directly compared quences in a form amenable to subsequent opera
with the last estimate . Relative movement during tions .
the last report period is immediately evident . - Run 06 - Event File Updating . The lasting his
( 2 ) Integrated Flow Charts . A second mode of torical record of previous transactions is on a mag
summary presentation is available in the develop netic tape known as the " event file . ” The purpose
ment of integrated flow charts . The major subsys of this run is to update this " file " so as to add ,
tems are joined together in this presentation to show delete , or change the information therein .
the entire FEM picture. Detail can be deleted from -Run 07 - Sequencing . The PERT analysis pro
the picture and thus only those activities that are cedure requires that the event flow chart be col
of paramount importance will be presented . lapsed into one straight line in such a way that all
This presentation can demonstrate another major connecting links point in one direction . Run 07
dimension of program status. Emphasis can be accomplishes this necessary sequencing activity .
given to the critical path and immediately highlight The program steps for achieving this sequencing
to overall management those subsystems which are highly novel . For this reason a detailed flow
should bear the greatest scrutiny . Exhibit D in chart and program are included in Appendix B.
Chapter I is illustrative of this type of presentation . -Run 08 - Slack and Probability Computations.
4. THE COMPUTER ANALYSIS AND OUTPUT .
This run accomplishes several objectives . First ,
the difference between the earliest and latest times
( 1 ) Capability of the ORC . The NORC installa for an event is computed . Then a series of steps
tion at Dahlgren is the nucleus of a modern high take place which compute probabilities of meeting
speed computing laboratory . The NORC is currently schedule . These probabilities are found by refer
the fastest computer operating in the world . It is from ence to a table of normal probabilities recorded in
two to five times faster than any other machine . It the NORC's memory .
can add or subtract 16- digit numbers at a rate of -Run 09 - Output Sort . This operation reorders the
approximately 22,000 a second . Multiplications of event information and structures it in a form that
16-digit numbers can be carried out at the rate of will facilitate appraisal of the output information .
14,000 a second . Run 10 - Output Edit . This run essentially se
lects from all the information that reposes the
The NORC system operates with eight high speed
magnetic tape storage units , each of which can tape that which will be subsequently printed . It al
transmit information within the computer at a rate of so places this selected information in a form which
71,400 digits per second . Input and output to the will be acceptable to the tape to card conversion
which follows .
computer is via magnetic tape . The laboratory has Certain other runs have been designed and are
a card -to - tape - to -card converter which will produce
or accept information in a punch card medium . In now in the process of programing and checkout.
addition, the Dahlgren facility has a large , well These additional runs will act to simplify and co
equipped tab card installation . ordinate the overall machine analysis. Modifications
of this kind are part of a continuing program analy
( 2 ) Machine Runs on the NORC . The PERT sis that will take advantage of the experience gained
procedure has been programed on the NORC in the in future machine runs .
form of ten subprograms or machine runs . The over
all program is broken down in this fashion accord ( 3 ) Accuracy Controls in the Program . The need
ing to the various operations that are accomplished for detailed controls on the flow of information out
during the cycle . Thus , ten distinct phases con side of the computer has already been noted . The
tribute to the outputs which have been discussed . arithmetic involved in the computer operation is
The various machine runs that comprise the checked within the operation of the computer itself.
program are described in detail in Appendix B. In In addition, good computer practice requires that
summary , however , the objectives of the runs can the program itself include checks for accuracy. The
be explained as follows : program checks that have been designed are of the
-Run 01 - Card to Tape Conversion . This run
.

following kinds.
takes incoming data punched in cards and translates - Internai Controls . Internal controls are built into
the information into a magnetic tape form . The data the program as a general check of consistency
can then be fed into the computer in the form of mag and reasonableness . The magnitudes of in
netized spots on plastic tape . put data are checked for proper relative size .
-Pun 02 - Data Transaction Conversion . This op A series of additional checks assure that internal
eration takes the incoming data and transforms the postings are made to the correct event “ accounts . "
time estimates into distribution parameters . The a , The efficiency of these checks should be under
m , and b ( optimistic , most likely , and pessimistic ) constant surveillance . Additional checks should be
values of the input are expressed as the mean and designed and incorporated into the program as they
variance of elapsed time distributions . appear desirable .
-Runs 03 , 04 , and 05 - Sort and lierge Operation .
- -

- Transaction and File Contrul Totals . These con

The first two runs sort the incoming event informa trols give positive evidence that the number of trans
tion into event code order . Run 03 deals with prec actions introduced into the machine are all processed .
ecessors to activities while Fun 04 orders the suc Trial totals are taken at various points during the
cessor events . Pun 05 merges the two ordered se runs. Machine stops occur whenever inequalities in

21
the totals are discovered . These totals must agree The following estimates were made on the as
with the log records kept by the SP control center. sumption that the entire FBM system would be cov
— Tape Label Control . The PERT system will gen- ered in a gradual build -up over the next seven
erat e many different record tapes. Each will be ap- months (through June 1959 ) . It is assumed that the
plicable to a given analysis . In addition , the NORC number of events in the computer event file for the
laboratory has thousands of similar tapes descriptive remainder of this fiscal year will average between
of the many programs they handle . It is essential 1,800 and 2,000 . It is anticipated that the average
that correct tapes be used in the machine analyses . number of monthly runs ( both regular and experi
In order to uniquely identify tapes , code words are mental ) will be approximately 60 .
magnetically written on the tapes . These code If these levels are realized , computer charges
words are checked automatically against the proper to SP will be at a $ 5,500 /month rate.
code which identifies ( and accompanies ) each batch In addition to the charges for machine time , some
of information submitted for analysis . program modification will be necessary . It is esti
( 4 ) Costs of the NORC Operation . SP computer mated that the necessary programing labor will come
expenses to date reflect two independent activities . to about $ 5,500 in total.
First is the detailed program information and check- Working on a basis of seven remaining months ( for
out on the current PERT system . The second is computing ) during this fiscal year , the total expected
representative of work done in discovering the in- requirement on SP is , therefore , between $ 45,000
trinsic validity of the current procedures . Both of and $ 50,000 . This is an estimate that should not be
these costs are one-time expenditures . exceeded in practice . Charges against a transfer
The costs incurred by SP during the initial pro- of funds to the Naval Weapons Laboratory will only
graming phase can be divided between machine time be made for actual labor and machine time usage .
and necessary programing labor. The total expendi- If the full FBM system is not covered by PERT by
ture from 1 July to 31 October 1958 has been $ 36,095 . 30 June 1959 , the charges will be less than the fore
( NORC machine time rates at $ 195 /hour are very going estimate .
low for a large scale machine . )
Actual programing on the PERT methodology has *

been responsible for $20,645 of this expenditure.

This fi gure represents approximately an equal divi
sion of funds between machine time and labor . The The previous chapters have described the work of
remaining funds , $ 15,450 , have been expended in the PERT team during Phase II . A concept has been
work on the validity check for the procedure . detailed - an installation plan has been formulated
In the future, NORC costs will tend to be di- and the first components have been covered by the
rectly proportional to the machine utilization at system . However , the job is not complete . The fol
Dahlgren . This in turn will generally be a function lowing chapter describes several important areas
of the rapidity with which the PERT system is which will need attention in order to fully realize the
installed . potential of the evaluation system .

IV . FUTURE ACTIVITY

Modification should be a continuous process as 2. EXTENSION TO THE ENTIRE FBM NETWORK .

it is applied to the PERT system . The basic analy- The PERT method has been designed for system
sis and installation structure has been set . Improve- application . Current plans call for making the en
ments will be discovered and should be incorporated tire application by 30 June 1959. It is most important
as their worth is established . Certain areas have that this be done efficiently . Serious oversights
already been recognized as in need of attention . can develop with less than a full implementation.
A whole series of critical events may never become
1. EXPERIMENTATION WITH OUTPUT FORMAT. evident if the entire system is not incorporated in
The present output format is simple and easily under the flow network .
stood . As SP works with the outputs , they un Extending the PERT system will involve many
doubtedly will find variations which will lend them- problems. Care must be exercised so as to prevent
selves to improved efficiency . The previous chapter the substitution of a quantity of information for the
has indicated the extent of the available information necessary quality . Haphazard installations can
evolved in the computer analysis . Sp 12 must work serve to impair the worth of other better considered
closely with the technical branches , providing them ones . It is recommended that contractors be care
with all of the information they can meaningfully fully coached and closely watched until their com
use . Only extensive experience will yield the best pliance is satisfactory .
form for presenting the output information . The SP Field Offices have an important role in
22
the installation and data-collection process . The unbiased estimates . As experience mounts , it
importance of their role must be made evident to should be possible to make statistical analyses that
them . Field office indoctrination should not be lim- will allow for calibrating the validity of a contrac
ited to the evaluation groups but appropriately tor's or individual's estimating ability . The basis
should include the technical groups . Time spent for such analysis will lie in the historical transac
in efficiently indoctrinating personnel in the field tion file within the NORC .
offices will yield a dividend in making for a ( 3 ) Sensitivity Analysis of the Present Method .
smoothly working system . The PERT networks frequently show situation
Extension of the PERT procedure to the full FBM where there are two or more sequences of activities
system will place a large load on the control center
within Sp12 . It is important that they quickly provide that are necessary in moving calculation
another. The present PERT from one eventtakesto that
for accomplishing the additional work . The magni sequence with the longest expected time as the only
tude of this effort should not be underestimated . constraint on the accomplishment of the second
Exhibit M , following this page , shows the estimated event. This procedure will underestimate the time
personnel requirements for a full installation within necessary to move between the two events . Occa
the current fiscal year . This exhibit shows that at sionally , a sequence with a relatively short time
the peak of the installation, Sp 12 will need approx expectation will slip significantly. The resultant
imately 13 people . After full installation, the re slippage can make the given sequence the major
quirements are somewhat less . For steady state constraint of the movement. The present procedure
operation after installation , it is estimated that 7 does not accommodate this possibility .
to 8 total personnel will be needed in 3 different
Mathematical analysis of this type of situation
position categories as follows : has been made . Current work is proceeding toward
Code estimating the extent to which the bias introduced
Description
by the present method underestimates predicted
A times . A separate report will be issued in which
Direct supervision , interpretation of re
sults , report writing, contractor, and SP
the results of the present work will be given . The
Field Office indoctrination . mathematics involved in obtaining an unbiased anal
B Contractor and SP Technical Liaison , ysis are complex . If the bias in the present compu
data handling in and out of SP , flow tation is small , the inaccuracies should be over
diagrams, monthly and quarterly reports , looked . If the bias is large , further work should be
interpretation . directed
of this toward
с NORC Liaison , computer input and out- C correcting the
report describes the original
situation.mathematical
Appendix
put handling, checking event nomenclature work leading to the accurate solution of the network
file, updating. problem .

It is important to recognize that failure to pro ( 4) Relation of PERT to Other Sp 12 Controls.

vide personnel in the number shown will result in a Broadly speaking , the PEPT computer program can
delay in achieving the installation objective . Time be fed information in sufficient fineness of detail
must be provided for adequat e training . The devel- to encompass all line -of-balance and PMP activity .
opment of the foregoing personnel estimates is However, a serious question is immediately apparent .
treated in Appendix D “ PERT Installation Person- Is this amount of detail necessary to effectively
nel Requirements . ” perform the job of SP management? Corollary to this
It is imperative that Sp 12 gear up their operation is the question " How many detailed production - type
so as to effectively accommodate information as it activities are needed in order to make the PERT
arrives . Present activity emphasizes field installa- outlooks meaningful ? Are not production activities
tion . When these installations are complete, data more definable and less subject to variance in their
will flow in relatively large volume . Future success achievement times ?
of the system can depend to a considerable measure These questions are important in structuring the
on the wisdom and efficiency with which initial in- networks and in decisions as to the number of de
puts are handled . tailed events to be included . To date , the philoso
phy of the PERT approach has been to move to a
3. REFINEMENTS IN ANALYSIS .
position halfway between the coverage detail of
( 1 ) Improved Computer Programing . Previous PAP and the line- of- balance charts . In individual
chapters of this report have highlighted the need branches , it will be found desirable to go into more
for making the computer program even more flex- or less detail as the situation requires or as is
ible than it currently is . Activity is now underway desired .
to accomplish certain improvements. These are One way of approaching the resolution to these
described in Appendix B. This work must be questions ( which we recommend to be answered only
closely monitored by SP so as to ensure the most as greater experience with the system is acquired )
effective processing . is to consider the NORC as a possible repository
( 2) Calibration of Estimatuts. Inferences from of all information relating to performance as sched
the PERT system depend upon estimates of the uled .
future . Individuals within the different contractors ' With this thought in mind, if it seems desirable
establishments will exhibit varying abilities to make to include more details such as the line-of-balance

23
 TOTAL REQUIREMENT
(INSTALLATION AND OPERATION )
In Man - Wks/Mo . "

Figures in parentheses ( ) indicate total equivalent number of personnel required

Total Position Category

Month Personnel A B С

1 ( 8.1 ) 7.98 ( 1.8) 18.87 ( 4.4 ) 8.25 ( 1.9)

)

2 (9.1) 8. 46 ( 1.95 ) 20.94 ( 4.8 ) 9.75 ( 2.3)

3 ( 10.0) 8.94 (2.06 ) 23.01 ( 5.3 ) 11.25 (2.6 )

4 ( 11.0) 9.42 ( 2.18 ) 25.08 ( 5.8 ) 12. 75 ( 3.0 )

5 ( 11.9 ) 9.90 (2.29) 27.15 ( 6.3) 14. 25 ( 3.3)

6 ( Target) ( 12.9) 10.38 ( 2.40) 29.22 ( 6.8) 15. 75 (37)

7 Operating rate ( 7.5 ) 3.36 ( 77)

.
14.49 ( 3.3) 10.50 (2.4)

* The above does not provide for training of installation staff.

EXHIBIT M

information in regard to any component , it is a very The foregoing sections discuss some of the
simple matter as far as the NORC is concerned . more important work that remains to be done before
One has merely to define and code the events , in- PERT is a smoothly functioning , accurate evalua
dicate their predecessors and successors , provide tion system . The concept of PERT is sound . Ini
the three time estimates (they may all be equal ) , tial installations have been successful . Top manage
and process as an input to the NORC analysis . ment has provided strong backing to the procedure.
Then , if reports are required riodically , the rou- The work has matured to a point beyor the " proj
tine reporting procedure described can be used to ect ” status , as is indicated by a sign ificant change
provide inputs to the NORC that will in turn provide in designation. PERT , as a code word , now signifies
printed output sheets showing status of these more “ Program Evaluation and Review Technique . ” Con
detailed events . tinued good work on the part of SP will result in a
As a display tool , we believe that line-of-balance system that will provide far- reaching insights into
has great utility at both the contractor and SP levels . the FBM development . Hopefully , the system will
We recommend that the question posed here - wheth- augment the Navy's currently operable policies and
er to use this detail in PERT - be answered by ex- aid in the successful and timely production of the
perimenting with such use and evaluating the expe- Fleet Ballistic Missile.
rience gained.

24
 APPENDIX A

INSTRUCTION MANUAL
and
SYSTEMS AND PROCEDURES
for
THE PROGRAM EVALUATION SYSTEM (PERT )

25
 TABLE OF CONTENTS

Page

I. INTRODUCTION 29

II. PURPOSE . . 29

III . INSTALLING THE SYSTEM .. 29

IV. MAINTAINING THE DATA FLOW . 37

V. UTILIZING OUTPUTS FOR MANAGEMENT DECISION MAKING . 42

27
 INDEX OF EXHIBITS IN APPENDIX A

Page

I. -
GLOSSARY (PERT) - FBM PROGRAM EVALUATION SYSTEM 30

II. -1 ORIGINAL FLOW CHART ( DUMMY ) ..... 32

II. -2 RE -DRAWN FLOW CHART ( DUMMY) 33

III. DATA FLOW DIAGRAM - PERT - ORIGINATING AND REVISION FLOW CHARTS
-
35

IV. (PERT) REPORT OF PROGRESS AND TIME INTERVAL ESTIMATES . 36

V. (PERT) REPORT OF PROGRESS AND TIME INTERVAL ESTIMATES . . 38

VI . ( PERT) REPORT OF PROGRESS AND TIME INTERVAL ESTIMATES ... 39

VII. DATA FLOW DIAGRAM - PERT - TRANSMITTING BIWEEKLY REPORTS

O
40

VIII . DATA FLOW DIAGRAM - PERT - REQUESTING ADDITIONAL INFORMATION ..... 41

1
IX. BIWEEKLY REPORT OUTPUT 43

28
 1. INTRODUCTION

The Program Evaluation System ( PERT ) is a method for evaluating progress to improve the planning of
a major research and development program . It is a tool for management control . Initially, the system has
been developed and applied to the Fleet Ballistic Missile (FEM ) Program .
The FBM program is very similar in form to any other type of plan since it is composed of a series of
activities or tasks scheduled in anticipated sequence culminating in the attainment of some final major ob
jective . Product performance and resource allocation operate as constraints on the time of accomplishment.
The problem of evaluating a research and development program is acknowledged as difficult by experi
enced research people because of the nature of the activities intellectual, pioneering , and unpredictable.
How can the Program Evaluation System (PERT) improve on a day- to- day comparison of progress with
planned schedule dates ? Basically ,the system can provide at a minimum ( 1 ) orderliness and consistency
in planning and evaluating , ( 2 ) automatic identification of possible future trouble spots as a result of failure
in one area, ( 3 ) speed in integrating progress evaluation , and ( 4 ) throughout these previous points, accurate
portrayal of the dynamic research situation .
II . PURPOSE

1. GENERAL DESCRIPTION . The purpose of this ( 4 ) Apply electronic data-processing equipment

Program Evaluation System ( PERT ) is to provide a (computer) to process and integrate these data.
means of handling the large and complex problem of ( 5 ) Establish information flow channels to as
rapid evaluation and reprograming in the FBM pro- semble progress and change data .
gram . The Systems and Procedures Manual provides 2. IMPORTANCE OF PROPER PROCEDURES .
a description of the methods used to obtain and Efficient operation of any evaluation system is de
process the required information necessary for the pendent upon timely receipt of accurate information .
effective operation of the system . This manual is The procedures outlined herein are designed to
intended as a guide for all personnel directly con- obtain the necessary information rapidly and with
cerned with operation of the system . minimum effort . In order to decrease the
Sections of the manual contain information to amount of work in volved and minimize the prob
cover the answers to the general questions : Who ? lems of installation and operation , the procedures
What ? When ? Where ? How ? Specifically , operation have been tailored to fit in with existing milestone
of the Program Evaluation System must develop progress reporting procedures wherever practicable .
along the following lines : The effective operation of the Program Evaluation
( 1 ) Select specific , identifiable events * which System is highly dependent on up - to -date computer
must occur along the planned progress toward some inputs and rapid dissemination of the outputs to
goal . management. The capability of the NORC computer
( 2) Sequence these events and establish the installation is utilized for processing the informa
relationships which exist among them . tion . The computer can process some 10,000 activ
( 3 ) Estimate the time and a measure of the ity times in less than half an hour.
variability of this time for the activity joining each
pair of events . * Exhibit I , on the following page , is a glossary of terms
that apply to the Program Evaluation System .

III . INSTALLING THE SYSTEM

1. PRELIMINARY MEETINGS. There can be no 2. ORIGINATING THE FLOW CHART -

THE
substitute for adequate preparation of all levels of VISIT OF THE EVALUATION TEAM .
management within the contractor organization ( 1 ) Selecting the Events . After the Evaluation
prior to beginning the actual installation . Probably Team has been properly set up and generally
the best approach to use is similar to the one fre briefed and trained , a meeting with the contractor
quently employed by SP management for originating assigned members of the team should be planned
milestone progress reporting procedures . Briefly , and held to select the events to be included on the
this involves briefing certain contractor personnel flow chart. It is possible that SP milestones used
in PERT methodology and outlining the procedures in existing progress reporting after examination and
to be used for gathering and transmitting essential necessary revision can become the events that are
data. This can be accomplished, in most cases , needed to form the basis for the flow chart .
at the time of a visit to the contractor's plant by a It is important to analyze each event description
well-trained team from SP with representatives from considered for use on the flow chart by asking ap
and arrangement by the SP Field Office. propriate questions in the following categories :
The importance of the acceptance and approval Category 1 - Does the event represent a definite
of management within the contractor organization discernible beginning or ending point of some activ
should not be underestimated. " Getting off on the ity or group of activities ? ( Such words as firm ,
right foot, " goes a long way toward making an ef finalize , freeze , ship , etc. are to be avoided . In
fective installation of the system . their place use such phrases as , “ design release
29
 GLOSSARY
( PERT )
FBM PROGRAM EVALUATION SYSTEM

ACTIVITY
An event is separated from other events by activities. An activity is a time - consuming element
in the development process. It is represented on a flow chart by an arrow . An activity cannot
be started until its preceding event has been accomplished. A succeeding event to an activity
cannot be accomplished until the activity is complete.
ACTIVITY TIMES
Estimates of the elapsed time necessary to complete an activity in a specified manner are
activity times. They are represented by estimates indicating:

Optimistic times
Most likely times
Pessimistic times

CRITICAL PATH
A critical path is that particular sequence of activities in a flow chart that comprise the most
rigorous time constraint in the accomplishment of the end event .

EVALUATION TEAM
The Evaluation Team ( Team) is descriptive of a group of designated individuals charged with
furthering the SP evaluation effort as it bears on a given sector of the FBM development.
Generally , a Team will be comprised of individuals representing the SP Technical and the
SP Evaluation divisions , SP Field Office , and the Contractor .

EVENT
A meaningful specified accomplishment in the FBM development program. An event should
be recognizable as a particular instant in time .

FLOW CHART OR PLAN

The sequenced diagrammatic representation of events and activities..
PRECEDING EVENT
See ACTIVITY

SLACK PATHS
Slack paths are sequences of activities that do not lie on the critical paths. Slack may exist
in varying amounts.

SUCCEEDING EVENT
See ACTIVITY

EXHIBITI

30
complete , " " production prototype design complete , " related and interdependent that unless 10 is
" approve for captive test , " " arrival of 1st flight completed the activity between them cannot
test vehicle at AFMTC , " etc. ) be started . Furthermore , it will be impossible
Categury 2 - Is the description detailed and com to reach the point in time represented by 22
plete ? Does it tell who does it ? Where and what until the activity represented by the arrow be
is done ? ( Be sure model descriptions are com tween them is completed . More specifically :
plete . Use specific FTV , CTV , and other avail an inspection of an assembly cannot be
able code numbers to name specific components , started until it is assembled , the assembly
subsystems , and systems . Use test and model cannot begin until the components are fabri
cated , etc.
numbers . )
cutegury 3 - Does the technical man who is to do
the estimating understand the events as beginning
and ending points of some clearly defined activity ? 10 22
Is he able to fix in his own mind what has to take
place at the completion of the preceding event be It is advantageous to start at the left of the
fore the immediately succeeding event is reached ? chart where some events may have been completed
( This category is used when interrogation of the already or scheduled for completion momentarily .
estimator takes place in obtaining time interval Some events are considered as having no preceding
estimates . ) event . These events start as of the time " 0 " or
" now . "
( 2 ) Laying Cut the Chart . After the events have Arrows leading into these events originate at
been properly selected , the next step is to place
the descriptions on a flow chart . This can be ac the “ O ” point . The connecting process is continued
across the chart to the extreme right until each
complished on a “ first cut " basis by drawing a
series of circles ( or boxes ) about 1-1 / 2 inches
event is properly related to all the other events
either
Exhibitdirectly or indirectly . Events 1 , 3 , and 5 of
across on a large sheet of paper. The event descrip- II, following this page, are events of this
tions are handwritten in the circles . ( They may be type .
abbreviated as long as a complete list of events The statement is sometimes made by the techni
and their description is maintained . ) Events are
placed vertically according to category (subcompo cal man , “ But this event does not have to be accom
nent on a component chart, subsystem on a system plished before we can spart the indicated subsequent
activity . "
chart, etc. ) and along the horizontal axis roughly The question to ask in this case is , “ Does the
according to schedule dates . It is best not to relationship as shown by the arrow , represent cur
place a permanent horizontal scale on the chart . If
rent planning within your group ? "
this is done , the estimator may use the divisions to ( 4) Estimating Elapsed
judge the planned amount of time scheduled to elapse nicalman, if available , whoTines. Thethesame
indicated tech
connec
between events and , thereby , introduce a bias into tions to be made between events should be con
his estimates .
sulted to provide estimates of elaps ed times and
At the conclusion of this step in the process , associated variabilities for the activities repre
the chart is nothing more than a series of circles
sented by the arrows between pairs of events . It is
with written descriptions in them . ( For the most
part , a sheet of paper about 24 by 36 inches should important that the technical man be fully aware of
the actions within his group which must take place
provide enough space for about 200 events with a between the events . Also , he must be able to pro
2 -year scale along the horizontal axis . ) vide three time estimates corresponding to three
( 3 ) Cunnecting Events . To aid in connecting different sets of conditions as indicated below :
events , the head of the technical group directly 1. Optimistic Time - The first estimate is an
engaged in doing the development work under con- " optimistic " one , in that it gives the best or shortest
sideration should be requested to provide qualified time . There is little hope of completing the activity
personnel . They should be persons of a high level in less than the optimistic time.
in the technical organization and yet should be 2. viost Likely Time - The “most likely ” time
capable of doing the development work themselves . estimate is that which would be given if only one
In other words , they should be aware of the overall time estimate was requested . It is the time that
planning as it applies to the area covered by the would occur most often if the activity was repeated
flow chart . under eract ly the same conditions many times . If
Technical personnel are to indicate the connec- many knowledgeable people were asked for the most
tians or arrows to be drawn between events ac likely time , the value given most often would form
cording to present plans and expectations. The the most likely time estimate .
criterion to use is , for example : 3. Pessimistic Time - If significantly worse
“According to present plans it is expected luck than usual occurs , the “ pessimistic time ” esti
that events numbered 10 and 22 are related mate indicates the longest time that the activity
and that event number 10 will have to be would take.
reached before event number 22 can be ac Time estimates should be entered on the flow
complished . Events number 10 and 22 are so charts along the wrows to which they apply . A good
31
 ORIGINAL
FLOW
CHART
D
)( UMMY

CHART
NUMBER
0271
10
OPTIMISTIC
LIKELY
MOST
PESSIMISTIC
AND
ESTIMATES
TIME

)(0,412 .
MOD ),21( 40
&o.
C
DX

DESCRIPTION
EVENT
3

( 3, 6, 9 )
OBJECTIVE
EVENT
.
CAPT
2ND
TEST
CIV
-2

( 9, 13 , 17 )
3),(12 CAPT
.IST ),2( 6 .
REL
DESGN

( .2, .4 , 1.5)
.
FLT
IST

32
TEST ,COMPL
DX
1-
1CIV
- 1
-
EX OBJECTIVE
EVENT
(4,

1 027000-100 )4
LL
EVENTS
:A
NOTE
HAVE
WHICH
6,

,3
8
)

UNDERSTOOD
ARE
NUMBER
PREFIX
NO (2,
027110
AN
HAVE
TO
.
PREFIX
( 14 , 14 , 20 )

EVENT
SUFFIX
NUMBER
FINAL IST
FLT .EX
1FLT
-
10
-
DX
(3.5, 5, 7)

( 10, 12 , 16 )
/000-200
02
(0, 5, 1)

(2,46) DELIVER
TO
(2, 2, 4)

PAD REL
.DES
1
-
DX .COMPL
1
DX

5
8

NOW
"TIME WEEKS
EXAIBIT
11-1
 CHART
FLOW
-DRAWN
RE
)(DUMMY

10
0271
NUMBER
CHART
:FOR
EXAMPLE
027110-003
NUMBER
EVENT

CODE
-
EVENT
MAJOR
HIGHER
FROM
NUMBER
CHART
LEVEL
3 027000-100

PATH
HEAVY
-
CRITICAL
2

33
6

027000-200

OBJECTIVE
EVENT
5

0 5 15 20 25 30 35 40

10
OR T
E
)(WEEKS
":NOW 2
-
II
EXHIBIT
form to use is illustrated in the original flow chart ( Begin ) Key word
of Exhibit II . When all three time estimates (opti ( Flight Test from
mistic, most likely and pessimistic ) are equal , the Flat Pad ) Complete description
activity should be analyzed carefully for possible (AFMTC ) Location , if necessary
errors in estimation . ( Propulsion s / s )
) Identification
(5 ) Identifying Events and Assigning code ( CX - 1 ) )
Nun: bers . In order to adequately identify events in ( 16 ) Prefix or sequence
a form amen able to computer analysis and to trans number
mit information about events in an unclassified
manner , it is necessary to assign code numbers to After all the events have been placed on the
the various events .
chart , suffix numbers are assigned : 001 to the first
The code number consists of two parts : a six event picked ( not in any prescribed order) , 002 to
digit prefix and a three digit suffix, separated by a the second , 003 to the third , etc. It is best to keep
hyphen . The prefix identifies the highest level an index of these events . The above example
would appear as follows :
chart number upon which the event appears . For ex
ample , most events on a component chart numbered INDEX
027110 would have this particular six -digit number
as a prefix. However , certain important events Other
which appear on a subsystem or system flow chart Identifi
would be numbered according to the topmost chart cation
upon which they appeared. Typically , the flow Code Number * Schedule
chart numbering would appear as one of the fol Number ( If Any ) Description Responsibility Date
lo wing examples :
027100-001 Begin Flight LMSD (--- )
Chart Test from
Number General Type Flat Pad ,
AFMTC
020000 System . Propulsion
s/s
027000 Subsystem (may contain 020000 numbers) CX - 1

027100 Component (may contain 020000 and * Any other number , such as chart and line number of a
027000 numbers ). PMP which should be retained.

027110 Subcomponent (may contain 020000,

027000, and 027100 numbers) .
( 6 ) Completing the Chart. The flow chart is to
be used as the basis for conveying the initial data
027111 Use of the last right hand column to Sp 12. In its original “ work sheet " form , it must
indicates a special chart. be classified because of the event description . If
the chart is redrawn with no time scale or dates
Note : It is possible to have a short- range system shown and the identification code numbers are
chart and number it 020001. Events on it , placed in the event boxes instead of the descrip
however, would be numbered 02000 if they tions , the chart can be transmitted unclassified .
appeared on the 02001 chart. However , an event index list is required in addition
to the flow chart to explain the meaning of each code
An index should be kept in SP probably by com- number .
puter liaison , of all high level chart numbers of a 3. TRANSMITTING INFORMATION .
general nature that are likely to appear on lower
level charts , i.e. , events that concern a combination ( 1 ) Using Flow Charts To Convey Information.
of two or more systems , subsystems , etc. A flight The chart should be redrawn so that it is readily
test , for example , would probably concern ballistic reproducible . If placed on tracing paper with Indi a
shell , flight control , guidance , and propulsion , etc.
ink lines , it can be reproduced by the Ozalid proc
When constructing a new flow chart or network , all ess . Original charts and reproducible masters
events of a general nature should be checked with should be kept by the contractor member of the
a copy of the index . When event descriptions are Program Evaluation Team and copies submitted
the same, numbering of the highest level chart must through the SP Field Office as per the data flow
take precedence . Notice the numbering of events chart in Exhibit III , following this page .
in Exhibit II . ( 2 ) Translating the Flow Chart to a Standard
Progress Formi. Information as presented on the
The suffix consists of three digits that are as flow chart cannot be fed directly into the computer .
signed sequentially as the chart is drawn . Descrip At some point in transmission , the flow chart has
tions of events are noted on the flow chart according to be converted into numerical data . Each activity
to the form and order illustrated by the following time and its corresponding boundary events are
dummy . translated into entries on the Report of Progress
34
 DATA FLOW DIAGRAM - PERT
ORIGINATING AND REVISING FLOW CHARTS

FLOW REQUESTS FOR REVISED

CHART CHANGES OR CHARTS
CORRECTIONS

SUB ESTIMATOR
CONTRACTOR ( Tochnical Man) O
( If axists PERT Team
Member
otherwise

primo)

APPROACH
RESPONSIBLE
MANAGEMENT
OFFICIAL

PRIME CONTRACTOR TEAM

( If No Sub - Contractor ,
Disregard This Line)

COGNIZANT NAVY FBM

FIELD OFFICE

SP 12 AND COMPUTER CENTER

SP 20 BRANCH

NAVY BURE AU
( As Required )

* RESPONSIBLE FOR ORIGINATING ( 5 COPIES) ( 2 COPIES) ( 5 COPIES)

COPIES OF FLOW CHART AND SUB
MITTING TO SP 12 .

KEY

DATA FLOW
USED IN PREPARATION OR
CONTACT & DISCUSSION
ORIGINATES
REVIEWS OR USES
FILES

MAY NOT BE NECESSARY

EXHIBIT III
35
 P
)( ERT
OF
REPORT
&
PROGRESS
TIME
ESTIMAT
INTERVAES
L

DUMMY

CLASSIFICATION
FILLED
)(WHEN
IN

CONTRACTOR
L
)OF
:(NOCATION
FROM
& AME :
TO .
NO
CONTRACT PERIOD
REPORT
FROM
TO

SP
12 CENTER
COMPUTER

OFFICE
FOR
ONLY
USE EVENT
IDENTIFICATIO
.
NO N TIME
E ST
.INT COMPLE
. DO
TION NOT
)(DO
IN
FILL
NOT PRECEDING SUCCEEDING DATE FILL
.
OPT MOST PESS
)(WKS LKLY W
() KS IN
REMARKS
)(WKS

(1) (2) (3) )

(B )
(C )
(D (E) )F( )
(G (4) )
(H

4 200 50 100-296
027 100-340
027 2 6 12 (NOT Effect
Proposed
of
Changes
Program
Entire
on
USED
)NEEDED

36
200 50 027100-297 027100-339 7 8

4 200 50 027100-298 027100-323 12 26 32

E
)( TC.

OFFICIAL
RESPONSIBLE
OF
TITLE
AND
SIGNATURE DATE
SIGNED

CLASSIFICATION
)(WHEN
IN
FILLED

IV
EXHIBIT
and Time Interval Estimates Form illustrated in runs , entries in these two columns will be supplied
Exhibit IV , following Exhibit III . The conversion by the individu als within SP Technical Branch or
should take place in the data flow process as soon the SP Field Office that originates the request for
as possible after the origination of the chart. A a special run . These entries will reflect any hypo
logical place for this routine clerical operation thetical or proposed changes in resources ( column
would be in the SP Field Office or in the contrac 2 ) and/or performance ( column 3 ). Entries should
tor's program plans or program analysis group . The be in the form of 3 digits representing per cent of
decision as to who does this will be made by the performance or resource allocation . Dropping per
Sp 12 representative at the time of installation . Re formance to 50 % of original plans would be indi-
quired copies of the completed forms are to be for cated by an 050 entry; similarly resource allocation
warded to SP for distribution and analysis as indi of 500 % ( 5 times original plans ) would be shown by
cated in Exhibit III. a 500 entry .
Column ( 4 ) is provided for the computer staff.
( 3 ) Coding the Standard Progress Forin . Certain Entries here translate a calendar date from column
sections of the Report of Progress and Time Inter (G ) into a code number understandable by the com
val Estimates Form are reserved for the coding sec puter. Tables will be provided the computer cleri
tion of the Computer Center or other authorized cal staff so that they can make the conversions .
groups . Entries in column ( 1 ) must be made as
follows :
* Leave this section out of manuals intended for use
solely by contractor personnel.
Code
Reason for Line Entry on Form Number 4. HANDLING OF ORIGINAL DATA .

New or original estimate of an ( 1 ) Revising the Original Flow Chart Drawing .

activity ( adding a connecting Computer analysis of the original data will deter
link) 1 mine for each event an earliest time or a time when
the event is expected to occur . After this is done ,
Pe- estimate of existing activity 2 revised flow charts will be drawn by Sp 12 Graphics
using earliest times for determining the positions
Completed activity 3 of various events along the horizontal axis . Copies
of these revised original charts will be transmitted
Static change - correction by to the cognizant SP Technical Branch and to the
computer staff estimator making the estimates . Any necessary
changes or corrections must be submitted to Sp 12
Deletion of an activity 5 Computer Center on the chart or on the proper form
Exhibit V , following this page.
Columns ( 2) and (3 ) are specifically set aside If there are corrections or changes to be made ,
for SP , FEM managers. Special computer runs can Sp 12 Graphics will revise the chart and notify the
be made when requested . In order to code computer original team of the changes ( see Exhibit III ) .
IV . MAINTAINING THE DATA FLOW

1. GENERATING THE BIWEEKLY REPORT . The 3. FULFILLING “ SPECIAL " REQUESTS FOR
form used with its attendant instructions is de SUPPLEMENTARY DATA. A "special " request
scribed in Exhibit Vi, on page 39. Transmission of for supplementary data may originate at any time
the form is outlined in Exhibit VII, on page 40. as a result of endeavors by SP Technical Manage
ment to develop new or revised plans and sche
2. REQUESTING ADDITIONAL INFOR TION dules . The method to follow in originating and
ON CRITICAL AREAS. As a result of computer transmitting these requests is outlined in the data
analysis , specific paths on the flow chart will show flow diagram of Exhibit VIII .
up as critical . At the end of each monthly period,
the contractor will be required to submit re- esti The process begins with the SP Technical
mates of all time intervals along the critical path . Branch and the SP Field Office writing up the re
The requests will be handled as set forth in the quest on a standard form ( Exhibit V ) and submitting
data flow diagrams of Exhibit VIII, on page 41 , and this to Sp 12. If more information is required on
will require either completion of a form similar to the form from the contractor, the partially completed
that illustrated in Exhibit V or a revision of a flow form is transmitted as shown in Exhibit VIII. Out
chart as provided by Sp 12. Requests from Sp 12 puts gener at ed as a result of such hypothetical or
may originate on the form mentioned , in which case , proposed changes are returned to the requesting
body .
it will be necessary for contractor personnel to
finish the entry for each partially completed line
and return the report to the Computer Center. Out 4. REVISING EXISTING FLOW CHARTS QUAR
puts will be distributed as illustrated in the data TERLY. Essentially the same procedure is fol
flow chart, Exhibit VIII. lowed in revising flow charts quarterly as is used
37
 )(PERT
PROGRESS
&
REPORT
OF
ESTIMATES
INTERVAL
TIME

DUMMY

CLASSIFICATION
)(WHEN
IN
FILLED

CONTRACTOR
L
)OF
:(NOCATION
FROM
& AME TO
: NO
CONTRACT PERIOD
REPORT
FROM
/T O

SALIFORNIA
,CLMSD
UNNYVALE 12
SP CONTRACT
I
(NSERT 9/30
TO
9/16
.)
NO

ONLY
USE
FOR
OFFICE .
NO
IDENTIFICATION
EVENT TIME
INT
.EST -I
COMPLE
DO
TION NOT
)(DO
IN
FILL
NOT MOST DATE FILL
PRECEDING SUCCEEDING .OPT PESS
LKLY IN REMARKS
)(WKS )(WKS
)(WKS

(1) (2) (3) )

(B )
(C )
(D (
)E (F) (
)G (4) )
(H

3 ."O 027110-001 9/22 12.6

** ovent
Completed

2 027110-003 027110-005 estimate

-
Ro
plans
in
change
of
because

38
5 027110-002 027110-004

5 027110-004 027110-005 dolotod

027110-004
#
Event

1 027110-002 027110-196 3 9
12

1 027110-196 110-005
027 4.5 6.5 8 added
*027110-196
Event 2
: /59
date
schedule

OFFICIAL
RESPONSIBLE
TITLE
OF
SIGNATURE
AND SIGNED
DATE

prodecossor
*No CLASSIFICATION
Personnel
Computer
or
12
Sp
by
in
Filled
** FILLED
)(WHEN
IN

V
EXHIBIT
 •No
prodocessor CLASSIFICATION
FILLED
)(WHEN
IN
Porsonnel
Computer
Sp
or
12
by
in
•Filled

EXHIBIT
V

)(PERT
PROGRESS
OF
REPORT
&
ESTIMATES
INTERVAL
TIME
DUMMY
CLASSIFICATION
IN
FILLED
)(WHEN

OF
:(NOCATION
FROM
L
)& AME
CONTRACTOR :
TO CONTRACT
NO
. PERIOD
REPORT
TO
FROM

CALIFORNIA
SUNNYVALE
–
LMSD 12
SP (INSERT
CONTRACT
.)
NO

ONLY
USE
OFFICE
FOR EVENT
.IDENTIFICATION
NO INT
TIME
.E ST |
cóMPLE-
DO
TION NOT
)(DO
IN
FILL
NOT PRECEDING SUCCEEDING .OPT MOST .
PESS DATE FILL
)(WKS LKLY W
() KS IN REMARKS
)(WKS

(1) )2( (3) )

(B )
(C )
(D (E) )
(F )
(G (4) )
(H
4

1 !
027110-001 027110-002 .1 .5 1 9/58

1 "0NOW
OR 027110-003 10 12 16 12/58

39
1 027110-002 027110-004 .2 .8 1.2 10/58

1 027110-004 027110-005 12 16 22 2/59

1 110-003
027 027110-005 2/59
3

1 110-005
027 02700-100 .2 .2 1 1/59 ( otice
02700
of
use
numbers
indicating
general
Nhigh
tlevel
, ype
event
)of
1 02700-100 027110-006 3 3 3 3/15/59

1 027110-009 10
10-0
0271 5 5 10 5/59

SIGNATURE
TITLE
AND
RESPONSIBLE
OF
OFFICIAL SIGNED
DATE

Contractor
than
other
by
in
illed
*FPersonnel
Column
D (Gnd
,EFailled
)f**
Contractor
by
out CLASSIFICATION
m
A ate
onth
considered
dear
aygiven
and
be
will
as )(WHEN
IN
FILLED
.
month
that
of
day
last
the
to
equal

EXHIBIT
VI

1
 DATA FLOW DIAGRAM - PERT -

TRANSMITTING BIWEEKLY REPORTS

PROGRESS CONTROL REPORTS

REPORT OR OUTPUTS

SUB ESTIMATOR
CONTRACTOR ( Technical Man)

( 1f oxists PERT Team

otherwise Member

prime)
RESPONSIBLE
MANAGEMENT
OFFICIAL

PRIME CONTRACTOR
( If No Sub - Contractor,
Disregard This Line)

COGNIZANT NAVY FBM

COPIE 2(

FIELD OFFICE
S)

SP 12 AND COMPUTER CENTER

-
1
SP 20 BRANCH

NAVY BUREAU
( As Required)
L
( 5 OR 6 COPIES) ( 4 OR 5 COPIES)

KEY
-

DATA FLOW
USED IN PREPARATION OR
CONTACT & DISCUSSION
ORIGINATES
REVIEWS OR USES
FILES -
MAY NOT BE NECESSARY .X

EXHIBIT VII

40
 DATA FLOW DIAGRAM - PERT
REQUESTING ADDITIONAL INFORMATION

COMPUTER
REQUESTS ANSWERS OUTPUTS

SUB ESTIMATOR
CONTRACTOR ( Technical Man)
( If exists -- PERT Team
Member
otherwise

prime)
RESPONSIBLE
MANAGEMENT
OFFICIAL

DI
PRIME CONTRACTOR

( If No Sub- Contractor,
Disregard This Line)

COGNIZANT NAVY FBM

FIELD OFFICE
Х Х

SP, 12 AND COMPUTER CENTER

SP 20 BRANCH

NAVY BUREAU
( As Required )

• EITHER SP 20 BRANCH OR ( 2 COPIES) ( 2 COPIES) ( 5 OR 6 COPIES)

SP 12 MAY ORIGINATE
REQUESTS

KEY

DATA FLOW
19

USED IN PREPARATION OR
CONTACT & DISCUSSION
ORIGINATES - O
REVIEWS OR USES
FILES

MAY NOT BE NECESSARY

EXHIBIT VIII

41
in setting them up originally. A team of repre ever necessary . All changes are communicated as
sentatives as described in Section III critically outlined in the flow chart of Exhibit III and are in
examines all events and times in the existing flow corporated as soon as available to Sp 12 in sub
chart. Revisions and re - estimates are made wher sequent computer runs .

V. UTILIZING OUTPUTS FOR

MANAGEMENT DECISION MAKING

1. INTERPRETING AND ANALYZING OUTPUTS . 2. DEVELOPING NEW OR REVISED PLANS. As

The output sheets as illustrated in Exhibit IX , fol a result of examining initial outputs on a particular
lowing this page, provide valuable information as component or subsystem , the Technical Branch of
to progress to date and anticipated future progress. SP may decide to develop and to test hypothetical
The TL - Te column indicates the amount of slack plans. Those events that are marked as probably
in a particular event . The presence of a large being reached ahead of schedule indicate possible
amount of positive slack in an event is an indi areas where resource trade - offs might be arranged .
cation of a place where resources might be avail Events along the critical ( zero slack ) path indicate
able for possible “ trade-offs . " A slack value ap possible areas for performance degradation or in
proaching zero would indicate that the event in creased resource application .
question is likely to be a potential trouble spot .
Critical paths are determined by highlighting (draw If the schedule for a major event is in jeopardy ,
ing in red or otherwise emphasizing ) those lines it may be possible to replan in a fashion that will
which pass through events with zero slack. Slippage improve the outlook of meeting schedule . This re
of an event along the zero slack path will cause a planning could take the form of altering the planned
corresponding slip in some major end objective . sequence of events , if that is possible. Such an
alteration could postpone certain activities to a
Monthly requests for additional data are an at later point in time. E.g. , it might be possible in
tempt to focus attention on relatively critical paths certain situ ations to forego some testing before a
and to audit a few specified events . Re - estimates flight. Thus the replanning would not force the
are requested for all those times that lie along the flight to await all of the preliminary tests that were
critical paths . This serves two purposes : ( 1 ) originally planned .
emphasisis placed on "tight" areas , and (2 ) esti
mates in these areas are marked for critical anal In addition to the changes already mentioned, it
ysis with an effort to produce more accurate data is possible to set up the computer so that it will
where it is needed most . develop a new schedule based on some arbitrary set
of criteria . This is described in detail in the Phase
Auditing seeks to improve the accuracy of the I report of the PERT project - available from Sp 12 .
flow chart by requesting estimates on a small
group of selected activities which differ from month Changes like those above can be indicated on a
to month. It is expected that all events on the form of the type presented in Exhibit IV , and the
chart will be covered at one time or another . computer will make the analysis within a few hours
after the origination of the request . The speed with
The Ts column is merely a conversion of the which the computer operates allows many hypothet
schedule date for an event to the coded date as uti ical situations to be posed and tested in a short
lized by the computer . The P , column is the result period of time . The difficult part of the process is
obtained by comparing on a probability basis the the translation of the physical situation envisioned
expected or earliest time ( TE ) for the occurrence of in the mind of the planner into the objective terms
an event with the schedule time ( Ts ) . The number which form the language of the computer. There
in the P, column indicates the probability of can be nothing hazy or indefinite about the informa
reaching the event on or ahead of schedule . Obvi tion or instructions that are fed into the computer
ously , if the P , value is low (below .05 ) , the prob if accurate and reliable results are to be obtained .
ability of meeting or beating the schedule is quite A computer liaison man will be available in Sp 12
remote. Values of about 0.5 are good , while values for consultation on problems of computer interpre
close to 1.0 are excellent . tation .

42
 PERT SYSTEM Biweekly Report Output
DATE 916 58 WEEK 11.8 PAGE

EVENT TE TT
T -TE TS Pi Scheduled Dato

027110-001 1.1 1.1 0 1.5 .65 9/26/58

027110-004 4.3 5.6 1.3 6.0 1.00 10/58

027110-005 29.3 29.3 0 25.0 , 00 + 2 / 59

027110-196 30.7 32.9 2.2 25.0 .00+ 2/59

EXHIBIT IX

43
 APPENDIX B

COMPUTER PROCEDURES

45
 TABLE OF CONTENTS

Page
I.

INTRODUCTION . 49

49
II . CHOICE OF THE COMPUTER FACILITY

III . PROVIDING A STRUCTURE FOR ANALYSIS ... 50

IV. THE PERT COMPUTER ANALYSIS 64

V. COMPUTER REQUISITES . . 79

VI. PRODUCTION OF TABLES ON NORC 85

VII . PROGRAM MODIFICATIONS 86

47
 INDEX OF EXHIBITS IN APPENDIX B

Page

I. COMPARISON OF COMPUTER FACILITIES AVAILABLE FOR PERT 51

II. PERT DATA PROCESSING FLOW CHART 52

III. INPUT CARD LAYOUT 54

IV. INPUT TAPE LAYOUT 56

V. UPDATING FLOW CHART - RUN # 06 57

VI. UPDATING ROUTINE 59

VII . DATA ITEM REFERENCE - FILE LAYOUT CROSS - REFERENCE . . . 62

VIII. EVENT FILE TAPE LAYOUT 63

IX. INFORMATION FLOW DIAGRAM - COMPUTATION SEQUENCE RUN ( RUN #07 ) . . . 65

X. PROGRAM FLOW CHART - SEQUENCING RUN # 07 67

XI . SEQUENCE ROUTINE . 69

XII. PROGRAM FLOW CHART - COMPUTATION RUN #08 . 73

XIII. COMPUTATION ROUTINE 75

XIV . EDITED OUTPUT TAPE LAYOUT 77

XV. OUTPUT CARD LAYOUT .. 78

XVI. DATA FLOW CHART - SP - 12 NORC CONTROL SYSTEM . 80

XVII. SP - 12 NORC DATA DISPATCH AND CONTROL SHEET 81

XVIII . NORC TAPE REGISTER .. 82

XIX . COST NPG - JULY 1 - OCTOBER 31 83

XX. PRELIMINARY ESTIMATES OF RUNNING TIME .. 84

XXI. MODIFIED SYSTEM FLOW CHART ..0 87

48
 I. INTRODUCTION

The advantages that will accrue from the PERT The PERT system analysis is being run on the
system depend on ability to speedily analyze a Naval Ordnance Research Calculator ( NORC ) at the
large mass of numerical data . The analysis itself Naval Proving Ground , Dahlgren , Virginia . The
involves many arithmetic operations . Human capac- planning for this application has been done by a
ity is unable to accurately cope with both the Booz , Allen & Hamilton staff specialist who , in
volume of data and complexity of analysis that is addition , has monitored the machine coding and
necessary for timely evaluation . Recognition of checkout . The actual machine coding has been
this fact ( which is conditioned by the immensity of carried out by programers from the Naval Proving
the FBM program ) inevitably led to the necessity Ground ( NPG ) installation .
of a computer analysis . During a large part of the installation period , the
A computer is a mechanism for following man- Special Projects Office was without an available
made rules with great speed and with great accuracy . computer liaison man . Consequently , it is partic
In the PERT application, the computer exists as a ularly necessary to provide an appropriate con
facilitating tool for technical managers . It is an tinuity of effort and transfer of knowledge and
adjunct that quickly summarizes for the manager a experience . This appendix will provide such con
group of facts that bear on the FBM program . These tinuity and transfer by a detailed description of all
facts will not make decisions for the manager . phases of the PERT computer installation .
They will , however , expand the base of knowledge The remainder of this appendix will trace the
which is necessary for wise decision . The com- history of the installation , note the problems en
puter processing of the PERT procedure will pro- countered and their solutions , and , in addition , will
vide important help in getting the necessary job highlight certain important considerations for con
done on time . tinuing computer application to PERT .
II . CHOICE OF THE COMPUTER FACILITY

1. GENERAL CONSIDERATIONS . The anticipated high speed memory , an electronic control section ,
size of the combined PERT networks for the com an arithmetic section , eight high speed magnetic
plete FBM system immediately directed attention to tape memory units , two line -at - a-time printers , and
the need for a high speed electronic computer . a console allowing manual direction of the com
Preliminary analyses of the magnitude of the job puter . Associated , but separate , equipment includes
indicated that complete coverage of the FBM sys a card -to-tape -to-card converter and a complete
tem would involve 5,000 to 10,000 events . Thus , a selection of card- processing equipment , 407 printers ,
high speed computing system is needed to maintain 514 ' reproducers , 077 collators , 082 sorters , and
accurate, detailed records of the events and to standard key - punch machines .
produce timely reports . Three modifications to the NORC system have
From the preliminary analysis , it was clear that been authorized and will be installed during the
the data-processing logic for PERT could be han next two years . The first modification ( now avail
dled by computer of the IBM 650 class . The need able ) is a cathode ray tube high speed photographic
for rapid reporting and possible expansion to more printer . ( Secondly , the memory will be expanded to
complex calculation led to the consideration of 20,000 words ( magnetic cores ) during the next year .
large computing systems . A further consideration Lastly , the Universal Data Transcriber ( UDT ) is
of importance was the availability of such systems being designed and built by the NPG staff . This is
within the Naval establishment . a unit containing a small scale computer designed
Consequently , a survey of nearby Navy estab specifically to handle the problems of converting
lishments was undertaken in order to choose that one input media to another . Generally , the UDT
facility which would best do the job from the stand will convert information in any media and code to
points of time and cost . any other media and code . Included in its capa
The equipment involved in the basic choice were bility is the transfer from one tape code to another ,
the Univac 1103- A at APL , Johns Hopkins Univer card-to-tape , tape -to-card , paper tape to magnetic
sity , and the IBM NORC at the Naval Proving tape , magnetic tape to paper tape , tape-to-printer ,
Ground . The comparative costs and operating capa card -to- printer , etc.
bilities of both systems are shown on Exhibit I ,
following this page . This analysis and some sub NORC carries out its calculations on decimal
jective consideration of the availability of person numbers arranged in 16 -digit groups called words ,
nel lead to the selection of the NORC installation . under control of a three address stored program . An
instruction word is arranged in five fields . Two
2. DESCRIPTION OF NORC . NORC is a high 2-digit fields ( P and Q) specify the form and opera
speed , stored-program computer . The computer tion of an instruction , and three 4 -digit fields ( R , S ,
proper consists of a 2,000-word cathode ray tube and T ) specify the high speed memory locations

49
required for execution of an instruction . A data Essentially NORC speed can be summarized by
word consists of 16 decimal digits which when used stating that NORC is at least two times , and in
in floating print operation consists of a 2 -digit co many cases four to five times , as fast as any other
efficient , a 1 -digit sign , and a 13 -digit mantissa . machine now in use , no matter how the problem is
In this manner , NORC can calculate using up to 13 considered .
significant figures in the range 10-30 to 10 + 30 . NORC is operated on three shifts daily , five days
The speed of the NORC is an important con a week . The primary shift is devoted primarily to
sideration as shown earlier ( see Exhibit I ) and can testing of programs , while the second and third
be summarized as below : shifts are utilized for production runs . Maintenance
Salient Speed Characteristics is scheduled and is conducted mainly during the
second and third shifts . Utilization of the equip
Memory Access Time 8 microseconds ment is in excess of 90% , when good time is ex
Instruction Time ( Average executive times pressed as a percentage of scheduled good time .
for arithmetic operations All maintenance and operating functions are per
including access time ) formed by the Navy .
Add -Subtract 46 microseconds
Multiplication 72 microseconds Having described both the facility that was
Division
chosen as well as the considerations that led to its
272 microseconds
choice , it is appropriate to associate the computer
Tape Time with the problem at hand . The computer processing
of the PERT information can be considered as two
Tape packing 510 digits / inch classes of activity . On one hand , a structure must
Tape transport time 140 inches /second be erected in the computer so as to accommodate
Tape information rate 71,400 digits/ second data and provide the framework within which anal
ysis can take place . Secondly , the analysis itself
The off-line card -to -tape -to -card converter ( CTC ) must be accomplished . The following two chapters
has a basic limiting speed of either reading or will describe the work involved in establishing an
punching cards at 100 cards / minute . efficient program in each of the foregoing classes .

III . PROVIDING A STRUCTURE FOR ANALYSIS

In order to provide a framework within which ( 2 ) Card Layout . Functionally , the card is laid
machine analysis can take place , two jobs must be out to meet the NORC operating requirements . As
performed . Initially , identifiable and meaningful
1 such , the NORC input card is constructed with four
data must be introduced into the machine . In addi 16 - digit words beginning at column 12 and end
tion , however , these data must be carefully placed ing in column 75 ( see Exhibit III , on page 54).
within the machine so as to expedite further anal Within these four 16-digit words , items of informa
ysis and subsequent outputs . tion can be laid out with little or no restriction on
The general flow of the data in the PERT sys the fields . However , to conserve on NORC operat
tem is symbolically portrayed in Exhibit II , on ing time and coding , it is advantageous to keep the
page 52. This flow chart shows the handling fields within each word and , when possible , to keep
of the stream of incoming information from the time each input transaction within one card . In develop
it is received by Sp 12 to its reappearance as a ing the system for mechanizing the evaluation sys
NORC output . tem , it was possible to satisfy both conditions . It
1. INPUTS TO THE COMPUTER is important to note that all NORC input must be
( 1 ) Key -Punch Procedure . Upon arrival at the either significant digits or zeros . Blanks or the
Computer Center , a batch of transactions will be absence of data are indicated by zeros .
assigned to the key- punch group . Here , input in ( 3 ) Card-to - Tape Conversion (Run 01 ) . After the
formation will be key- punched in cards while main cards have been punched and verified , they enter
taining batch identity . After key - punching , the into what has been designated as computer Run 01 .
cards will be key- verified . The key- punch time can The card -to -tape conversion is not a programed run
be estimated at 100 cards per hour , and key - verify in the sense that a stored program is necessary to
ing can be estimated at 100 cards per hour . The direct the operation . The CTC is an auxiliary unit
following are estimates for the weekly key -punch in the NORC Computer Center which functions
ing and key-verifying times for given sizes of either to produce punched cards from magnetic tape
event files . or to produce standard NORC magnetic tapes from
Key - Punch properly prepared cards . The CTC unit contains its
Number of and Verify own 100 -word cathode ray tube memory , a separate
Events Cards / Week Time (Hours ) magnetic tape unit and a modified IBM 519 reproduc
ing punch , which acts as a card punch and card
500 100 2.0 reader . The unit operates at a rate of 100 cards per
1,000 150 3.0 minute either punching or reading cards .
2,500 375 7.5 The 100 - word memory is the heart of the CTC
5,000 600 12.0 operation , as it acts as a buffer or speed exchanger
10,000 1,000 20.0 between the high speed , high density magnetic tape

50
 COMPARISON OF COMPUTER FACILITIES
AVAILABLE FOR PERT

Available
NORC 1103- A Type

Cost

Machine Tiine $ 195 /hr. $ 220 /hr.

Efficiency 3 5 1

Relative Cost
(Cost/unit of work) $ 40- $ 65 /hr. $220/hr .

Programming Cost Labor and Station O.H. Fixed Cost( 1)

Flexibility

Staff Large Small

Experience 10 Years 1 2 Years

Peripheral Equipment Extensive Limited ( 2 )

Interchangeability None Extensive

Navy Control Direct ( BuOrd ) Indirect

( 1 ) Approximately equal to NORC Labor and Overhead .

( 2) No off-line card - to - tape , or tape -to -card equipment available .

Serious cost limitation.

EXHIBIT I

51
 PERT DATA PROCESSING FLOW CHART

( PART 1 )

NEW COMPLETED
ESTIMATES EVENTS

NEW STATIC
DATA
EVENTS
CHANGES

KEY
PUNCH

TRANSACTION
CARDS

01

CARD TO
TAPE RUN

TRANS.
TAPE

CONVERT 02
o, m, b
TO
to , o'te

P S ON LINE
TAPE TAPE PRINT

03 04
CONTROLS
SORT PRED SORT SUC. REGISTER

P S
TAPE TAPE

MERGE
P TAPE WITH STAPE

TRANS .
TAPE

EXHIBIT II

52
 PERT DATA PROCESSING FLOW CHART
( PART 2 )

TRANS. ACTIVE
TAPE FILE

06

UPDATE FILE

1
ON LINE
PRINT
COMP
EVENT
FILE
HISTORY
NEW
ACTIVE
FILE
1
OFF 07
ERROR LOGICAL
REGISTER LINE SE QUENCE
PRINT
CONTROLS RUN

COMPLETED ACTIVE ON LINE

EVENTS FILE PRINT

08
COMPUTING CONTROLS
RUN REGISTER

ON LINE ALL
PRINT EVENTS
TAPE

09
CONTROLS SPECIAL SORT / EXTRACT
REGISTER EVENTS TO NATURAL
REPORT ORDER

ALL
EVENTS

10

EDIT OUTPUT
HOD

SPECIAL ALL
EVENTS EVENTS

OFFLINE OFF LINE

PRINT PRINT

SPECIAL ALL EVENTS

EVENTS
REPORT LISTING

EXHIBIT II

53
 LAYOUT
CARD
INPUT

BATCH KEYPUNCH PREDECESSOR SUCCESSOR SCHEDULE ZEROS .

PERT
NUMBER DATE CODE ZEROS CODE DATE
NWO

O OMN
son

Omn
Doomzo
.YRAY

(
m
)

)
a
(
D

OR
.MO

)
b
(
amm

URO VOOW
voi tuoidu

--3-00
J - xWJ

a woo3 zow
ZO - mo300

moon
-n - 3 - uumo

EW
5678910
18
17
16
15
14
13
12
11 26 27 34
33 44
43
42 48
47 52
51 56
55 59 60 65 66 77
76
75 80

54
IV

11

III
EXHIBIT
and the relatively slow speed , low density punched rected . Also the test ože > 100.0 is performed and,
cards. In the card-to-ta pe operation the unit reads although the transaction is not rejected , a print-out
24 PERT system input cards into the memory . This is made for further outside verification of this large
group is preceded by a 1 -word card , indicating the value of ože. The values for te and ož are com
beginning of the group or block and is followed by a puted and stored in the floating point mode using a
1 -word card indicating the end of the block . When full NORC word . In addition to the printed outputs
the memory is loaded with the 100-word information , noted above , the program prints out the input and
it is read out of memory into magnetic tape . Thus output labels for the tapes themselves .
the 24 IBM cards are compressed into 3.33 inches of The transaction data conversion program prepares
tape . In the card-to-tape processing the CTC unit two output tapes , one to be used in setting up ( or
provides a check dig it for each word . This is later updating ) the activity times for the predecessor
used as a part of the internal automatic checking event and the other to be used in setting up (or up
operation which is part of the basic NORC design . dating ) the successor event . A code identifies
The large difference in speed between the card and which of the two events is being handled (successor
tape functions allows the cards to be read continu- or predecessor ) . A six is used in the case of the
ously while the tape transfer takes place between predecessor event updating and a seven for the
the reading and movement cycles of the 26th and successor . The successor event identifies the file
27th cards , and then again between the 52nd and event and carries all the static data for the event
53rd card , etc. file .
As a result of this operation the PERT input ( 2 ) Sort and Merge Transaction Tape (Runs 03,
tape is created , and a detailed section of it is
.
04 , 05 ). Runs 03 , 04 , and 05 are shown on the flow
shown in Exhibit IV, on page 56. The tape is chart (Exhibit II ) as separate logical functions and
organized in 100 word blocks or sub -blocks. are separate runs at the present time . However ,
2. POSITIONING THE INPUTS WITHIN THE they will be combined into a single run in the near
COMPUTER . The magnetic tape output of the future and can be considered as a single function .
card -to -tape conversion is the vehicle for actually The purpose of these runs is to sort the con
introducing information into the computer . How- verted transaction tapes in event order sequence
ever , before the input information is used in an and then merge them . The predecessor updating
analysis run , it is necessary to precisely locate it transactions are sorted into event order on Run 03 .
and change it to a form most amenable to the sub- Then the successor updating transactions are
sequent analysis . In order to accomplish this , sorted into event order on Run 04. Following this ,
several computer runs are made . The result of these the two sorted tapes are used as the input tapes for
runs is the establishment of a basic event file , the the merge run , Run 05. The two tapes are merged ,
final repository of the basic inputs ( see Exhibit II ) . maintaining event order with the predecessor events
going on to the output tape first in the case of
( 1 ) Transaction Data Conversion (Run 02). The identical event codes . The resulting output tape
input to Run 02 , the transaction data conversion ( in event order ) is arranged in 196 word blocks with
run , is the tape from the CTC , Run 01. The input is
transaction messages seven words in length. Each
arranged in 100 word blocks , 24 ' transactions to a
.
transaction with a predecessor event code has
block . Each transaction is arranged in four word sufficient information to update the successor activ
messages , see Exhibit IV . Transactions are not
ordered in any sense for this run . ity of the event indicated by the predecessor code .
Each transaction with a successor event code has
The primary purpose for this run is to convert the information necessary to update the predecessor
the three time estimates -a , m , and b-into values of activity of the successor event . In addition , such
te and 02. and to set up two output ta pes for up information as schedule date , Ts,' completion date
dating predecessors and successors .
( in week code form ) , resource , and performance rate
In converting a , m and b to te and o?, the codes are associated with the successor event
program uses formulae of the following form .
transaction .
te = K, m + K, (a + b ) The output of the three runs is tape with
2
b - a
02. = 6 K3 ordered lists of successor and predecessor events
together with associated information . An additional
run is now necessary before the event file contains
The three parameters K1 , K2 , and K3 are in- the input data . This run ( updating) is programed to
serted in the program at the time the run is set up accommodate either an original file set up or the
and can thus be changed at will if needed . At the altering of an existent file of events .
present time the values used are as follows :
K , 0.6666666666667 ( 3 ) Event File Updating (Run 06 ). The file up
K2 = 0.1666666666667 dating run is designed to add , delete , or change
K3 = 6.0 information in the event files . Two input tapes are
The run 02 also tests the relationship : required , first the even file tape and , second , the
.

a < m <b transaction tape . Both of these are in event order .

and rejects all transactions where this inequality is This program will handle five and only five trans
not met . These are printed out so they may be cor- actions .

55
 INPUT TAPE LAYOUT

3
WORD
1 2 3 10 12 13 14 15 17
# 4 5 6 7 8 9 11 16
EVENT
CARD RES. RATE PREFRATE PREDECESSOR CODE
CODE ✓
1 0
х х х O x x X - x x х х х O х
RPT .
SUCCESSOR CODE
CODE
N

0 0 0 0 o o Тх х X - X X х — x x х х х

COMPLETION OR
OPTIMISTIC EST . ( a) MOST LIKELY EST . ( m ) PESSIMISTIC EST . ( b) SCHEDULE DATE ✓
3
x x Х х х x , X х х х х X х x x Х X

4
SCHEDULE DATE ( CAL . ) ✓ )
X x - x x - x Х 0 o o o o o o o o 1 х

O
EVENT

EXHIBIT IV

-Transaction Code 1. Code 1 indicates that the repeated . If the transaction is equal to file event
]
V. EST . (
transaction message contains a new activity which code , the program branches into the updating sub NO
is not now in the file . An option switch described routine . The file event message at this point is
later also allows this type of transaction to set up a extracted from the read -in area and is placed in a
new file from blank tape . Transactions with Code 1 working memory position . The transaction code and
add a new activity to the file event message in the predecessor -successor update code are tested .
either the predecessor or successor group . In addi A new activity code ( Code 1 ) will cause a new
tion , transaction Code 1 will set up new information successor or predecessor activity code and the
in the performance and resource rate fields , as well associated activity time estimates to be added to
as schedule information . the event file . A re- estimate code ( Code 2 ) will COMP
- Transaction Code 2. Transaction Code 2 indi cause a new set of activity time estimates to re CTIVITY

cates that the transaction message contains data to place the existing estimates in the file event . .. NO

be used to add a re- estimate of the activity time for successor or predecessor activities . A completion
the indicated predecessor or successor activity of code ( Code 3 ) will cause a predecessor -successor E " DATE
the event specified . Only transactions with Codes activity to be removed from the file event record and TATIC

1 and 2 contain te and on data . the successor or predecessor index to be adjusted . NO

-Transaction Code 3. Transaction Code 3 ind i In addition , the predecessor index will be tested
cates that the transaction message contains informa after modification . If it is zero , the completion date
tion that an activity has been completed . The trans will be inserted in the completion date field . Fol
action removes the activity from the file and inserts lowing this , the entire event file record will be read
a completion date in the file when the last prede out into the completed event tape . A static change
cessor is completed . code ( Code 4 ) will cause the file event static data
LETE
-Transaction Code 4. Transaction Code 4 indi to be replaced with the static data indicated in the TIVI TY
cates that the transaction message contains data to transaction message . Note the only static data
NO
be used in making changes in the static portion of change permissible is a change in schedule date ,
an event record . The static data is contained in the report code of Ts . A deleted activity code (Code 5 ) RINT OUT
first seven words of the file . A static data change will cause the indicated predecessor or successor
will affect only schedule dates , and report codes . activity to be removed from the event file record . No
-Transaction Code 5. Transaction Code 5 indi test is made of the predecessor index and no write
cates that the transaction message contains data to out of deleted events is made .
be used in deleting an activity from the predecessor The cases for the transaction code's being 105
event ( a successor activity ) and from the successor greater than or equal to the file event code have
event ( a predecessor activity ) . " Delete activity " been described . If the transaction code is less than
transactions are similar to completion transactions the file event code , the transaction code is tested .
in overall function except that they do not adjust If the transaction code is a new transaction code
the completion date . (Code 1 ) and the updating code (predecessor
The program operates in the following manner : successor ) indicates a successor event to be up
The first block on the file tape and the first block dated , a new event is constructed in working stor
on the transaction tape are read into memory . The age and then read out into the output tape block . If
first transaction event code and the event code for any other code combination is identified , then the
the first event in the file are compared through a transaction is rejected and the transaction data
subtraction . If the transaction code is larger , the printed on the error control register . When a trans
next event in the file is selected and the process action is rejected , the next transaction is extracted

56
 UPDATING FLOW CHART
RUN # 06
230 700 259

B -B MAKE UP EVENT STORE MAKE UP STATIC WDS.

EVENT A STORE OF EVENT 7 WDS .
AL SO PRED,
EVENT CODE SUC, ANDO WOS.
FOR EVENT AND ?WE SET EXIT -26
0

204

SET EXIT # 26
SET EXIT 26. SET
EXIT IN BOX 204 TO
BOX 700 OR 750

724
55 66
SENSE OUTPUT COMP .
200
63

0
700 750

SET EXIT 25 AND 17 SET EXIT 16 AND # 26

W EVENT

728
YES
EV . EST SENSE : TRANS. - PRED STORE STATIC PART OF
OR SUC = 0
EVENT OUTPUT
NO
NO

REV, EST FOR . AND

PRINT OUT
oi. FOR PRED. AND
STORE PRED CODES
SUC. LIST OF EVENT
SET EXIT # 26

105 724
STORE SUC. CODES
COMP
CTIVITY
NO

# 26
105 EXIT
NEW DATA YES
STATIC CHANGES IN
EVENT
STATIC © # 25

NO 730

MAKE UP WRITE INST .

10191 XXXX XXXX 0001
724 VERIFY 1 . BLOCK 4

YES
SENSE : TRANS - OD 18
LETE
( TIVITY OR SUC -0 EXIT 728

NO 0 # 17

RINT OUT DELETE PRED OR SUC

ACTIVITY FROM EVENT 105

105 724

EXHIBIT V
 UPDATING ROUTINE pago 1
-02 91 0051 0056 0000 09 96 0008 1727 0001 00 6A 0000 0000 0006 00 60 0000 0000 0009
00 00 0000 0000 0000 00 61 0000 0000 0001 00 00 0000 0000 3951 02 91 0051 0056 0000

-02 91 0051 1770 0001 00 00 0000 0000 0000 no 54 0000 0000 Vvoo 02 94 1031 1031 0000
08 68 0011 0000 1687 00 61 0000 0000 0088 00 40 1553 1578 1578 00 70 1031 0000 0016
on 61 noon onnn Ono on 60 1553 On8 0000 00 40 157 1583 1578 01 91 1578 15 7A 0000
08 60 0019 0000 1704 no 74 0000 0000 0027 03 94 1032 1032 0000 OA 6A 0022 0000 1687
00 40 1979 1552 1579 00 60 0000 0000 0000 00 70 1579 1032 0027 00 61 0000 0000 0000
00 80 1032 0001 0000 on 80 1578 0002 0000 00 80 15BA0003 0000 00 BO Onoo On 04 0000
00 80 0000 0005 0000 on 80 0000 0006 0000 00 3 0000 0007 0000 02 94 0629 0000 0001
OB 68 00 35 0000 1687 00 67 0000 0000 0599 00 60 0000 1554 0000 04 42 0000 1585 1555
08 42 1554 1586 0000 00 41 1555 0000 1556 no 40 1587 Onon 1556 00 40 0034 1587 00 34
00 51 0000 0000 0000 00 60 0000 0000 0000 00 74 0000 0000 0057 03 94 0830 0000 0001
08 68 0047 0000 1687 00 67 0000 0000 0607 00 60 0000 1558 0000 04 42 0000 1585 1555
08 42 1558 1586 0000 00 41 1555 0000 1559 00 40 1587 0000 1559 00 52 0000 0000 0000
00 40 0046 1587 0046 00 60 0000 0000 0000 00 41 1556 1557 0000 00 63 0063 0060 0000
00 61 0000 0000 0000 00 60 1669 0065 0000 00 60 1588 1557 0000 00 60 1589 0044 0034
00 60 4629 1033 0000 00 40 0063 1590 1063 n4 5A 0001 0007 0063 96 60 1591 Onno 0000
00 40 0065 0000 0065 00 60 1592 0063 0000 00 40 1557 1591 1557 00 42 1036 1593 1560
00 70 1593 00 00 00 74 00 70 1560 1594 0081 00 61 0000 0000 0000 00 60 1607 0146 0000
on 60 1614 0254 0000 on 60 1614 0291 0000 00 60 0341 0329 0000 00 60 1680 0422 0000
00 60 1666 0414 0000 00 60 1668 0342 0521 00 60 1608 0146 0000 00 60 1633 0254 0000
00 60 1635 0291 0000 00 60 161 4 0329 0000 00 60 1656 0422 0000 00 60 1614 0414 0000
00 60 1614 0342 0551 00 80 1553 On7 0000 00 80 1031 0001 0000 00 80 1588 0003 0000
00 BO 0000 0004 0000 00 80 0000 0005 0000 no 80 1552 0006 0000 00 81 15AA 0002 0013
00 60 0000 0000 0000 00 41 1559 1561 0000 00 63 0102 0099 0000 00 61 000 0000 0000
00 60 1588 1561 0000 00 60 1595 0056 0000 00 60 1601 0124 0046 00 60 1588 1563 0000
00 60 1588 1564 0000 00 60 6836 1562 0000 02 42 1562 1582 1563 00 42 1562 1596 1564
00 40 1563 1597 0000 50 24 1598 0000 0000 no 34 1599 0000 1563 00 40 1564 1597 0000
50 24 1598 0000 0000 00 34 1599 0000 1564 00 60 1591 1565 0000 00 72 1563 1588 0116
00 60 1670 0130 0118 00 72 1564 1588 0118 00 60 1671 0138 0000 00 40 1563 1564 0000
00 40 1565 0000 1565 00 59 0000 0000 0000 00 42 0000 1585 1575 00 60 6830 1041 0000
00 40 0122 1590 0122 06 58 0001 0007 0122 00 60 1600 0122 0000 00 42 0124 1585 0132
96 60 1563 0000 0000 00 40 0132 0000 0000 00 40 1602 0000 0132 00 60 6830 1048 0000
00 40 01 30 1590 0130 00 00 0000 0000 0000 00 60 1603 0130 0000 00 42 0132 15 A5 0140
96 60 1564 0000 0000 00 400140 0000 0000 00 40 1604 0000 0140 00 60 6830 1148 0000
00 40 0138 1590 01 3A 00 00 0000 0000 0000 00 60 1605 0138 0000 96 60 1565 0000 0000
00 400124 onoo 0124 00 40 1561 1565 1561 15 60 1033 1569 0000 00 42 1033 1606 1568
00 42 1042 1606 1567 00 70 1568 1567 0150 00 60 1637 0224 0163 00 41 1568 1567 0000
00 63 0152 0235 0411 00 70 1587 1569 0161 00 80 1033 0001 0000 00 80 1034 0002 0000
00 80 1035 0003 0000 00 80 1036 0004 0000 00 80 1037 0005 0000 00 80 1038 0006 0000
00 83 1039 0007 0000 00 60 1638 0095 0057 00 60 1637 0224 0163 00 60 0000 0000 0000
00 60 1591 1565 0000 00 40 1563 1564 0000 00 40 1565 0000 1565 00 40 1570 1565 0000
00 41 1609 0000 000 on 63 0172 0169 0171 00 60 0224 0234 0000 00 60 1610 0224 0172

UPDATING ROUTINE page 2

00 60 1611 0234 0225 00 54 0000 0000 0000 97 60 1041 0000 0000 03 60 0000 1041 0000
02 42 1047 1582 0000 88 34 1623 0000 0000 87 60 0000 0000 0000 00 40 1041 0000 1041
00 42 1047 1596 0000 88 34 1623 0000 0000 87 60 0000 0000 0000 00 40 1041 0000 1041
87 60 1591 0000 0000 00 40 1041 0000 1041 13 60 0000 1565 0000 00 60 1041 9348 0000
00 40 0186 1612 0186 00 60 0000 0000 0000 08 58 0001 0007 0186 00 60 1613 0186 0000
00 60 1614 0172 0000 00 72 1563 1588 0194 00 60 0000 0000 0203 00 42 0189 1585 0202
96 60 1563 0000 0000 on 40 1616 0000 0000 00 40 0 202 0000 0202 00 60 104A0348 0000
00 40 01 98 1612 0198 on 70 9348 1593 0588 00 40 1577 1587 1577 00 00 0000 0000 0000
00 60 1617 0198 0000 00 60 1588 1577 0000 00 72 1564 158A 0207 00 60 0000 0000 0217
00 59 0000 0000 0000 96 42 0000 1582 0216 96 60 1564 0000 0000 00 40 1618 0000 0000
00 40 0216 0000 0216 00 60 1148 9348 0000 00 40 0212 1612 0212 00 70 934 A 1593 0594
00 40 1577 1587 1577 00 00 0000 0000 0000 no 60 1619 0212 0000 00 60 15AA 1577 onoo
96 60 1565 0000 0000 00 400189 0000 0189 no 60 1588 1563 0000 00 60 158 1564 0000
00 40 1570 1565 1570 00 00 0000 0000 0000 96 60 1570 0000 0000 00 41 0000 1500 1570
00 40 1620 0000 022 A 00 00 0000 0000 0000 08 60 0229 0000 1704 00 40 1620 1587 1620
00 60 1621 0189 0000 00 60 1615 0172 0000 00 60 1588 1570 0000 00 00 0000 0000 0000
00 70 1569 1587 0422 00 70 1569 1622 0254 00 70 1569 1623 0342 00 70 1569 1624 0329
00 70 1569 1625 0291 00 80 1578 0001 0000 00 80 1584 0002 0000 94 80 1553 0003 0000
00 80 1588 0004 0000 00 80 0000 0005 0000 00 80 0000 0006 0000 00 81 0000 0007 0153
00 80 1578 000 0000 00 80 1568 0002 0000 00 80 1043 0003 0000 00 80 1044 0004 0000
00 80 1045 0005 0000 00 80 1046 0006 0000 0081 1036 0007 0377 0000 0000 0000 0000
00 42 10 36 1606 1568 00 42 1148 1606 0000 no 70 1568 0000 0263 00 40 0256 1626 0256
00 40 1971 1623 1571 00 41 1564 0000 0000 00 63 0256 0000 0000 00 60 1683 0410 0404
00 42 10 36 1593 0000 on 41 1036 0000 1036 04 42 0256 1586 0000 00 40 1590 000 0000
00 00 0000 0000 0000 00 40 0262 1629 0270 00 00 0000 0000 0000
00 40 1628 on 0268
00 60 1630 0256 0000 00 60 1588 1571 0000 00 60 163A 0095 0057 00 42 1033 1606 1568
00 42 1048 1606 0000 00 70 1568 0000 0282 00 40 0275 1626 0275 00 40 1623 1571 1571
00 41 1563 0000 0000 00 60 1684 0410 0404 04 42 0275 1586 0000
00 63 0275 0000 0000
00 40 1590 0000 0000 on 40 1631 0000 1285 0000 0000 0000 0000 on 40 0 285 1629 0287
00 00 0000 0000 0000 00 60 1638 00 95 00 57
00 60 1632 0275 0000 00 60 158A 1571 0000
00 42 114 R 1606 0000 00 70 1568 0000 0300
00 00 0000 0000 0000 00 42 1036 1606 1568
00 40 0293 00 40 1571 1623 1571 00 41 1564 0000 0000 00 63 0293 0000 0000
1626 0293
04 42 0293 1586 0000 00 40 16 34 0000 0302 0000 0000 0000 0000
00 60 1685 0410 0404
00 40 0302 1590 0304 0000 0000 0000 0000 00 40 0304 1590 0306 00 00 0000 0000 0000
00 41 1047 1587 1047 00 42 1033 1606 1568
00 60 1630 0293 0000 no 60 1588 1571 0290 162 ? 1571
00 40 1571
00 42 1048 1676 0000 on 70 1568 Ono 0318 no 40 03111626 0311
00 63 0211 0000 0000 00 60 1686 04100404 9A 60 1587 0000 0000
00 41 1563 0000 0000 00 0000 0000 0000
0041 1047 0000 1047 04 42 0311 1586 0000 00 40 1634 0000 0322 0000 0000 0000
00 00 0000 0000 0000 00 40 0324 1590 0326 00 00
00 40 0322 1590 0324
00 60 1632 0311 0000 00 60 1588 1571 0290 00 00 0000 0000 0000 01 60 1042 0000 0000
00 42 1036 1587 0000 96 60 1047 0000 0000
99 60 0000 1042 0000 00 40 1042 0000 1042
88 42 1039 1582 0000 00 42 1041 1584 1041
04 60 0000 1047 0000 00 40 1047 0000 1047
00 00 0000 0000 0000
10 42 1029 1580 0000 00 40 1041 0000 1041 00 60 0000 0000 0057
00 400344 1626 0344
00 42 1033 1606 1568 00 42 1048 1606 0000 00 70 1568 0000 0351
00 60 1632 0344 0402
00 40 1571 1623 1571 00 41 1563 0000 0000 0 63 0344 0000 0000

EXHIBIT VI

59
 UPDATING ROUTIIB page 3
04 42 0344 1586 000 00 40 1634 0000 0353 00 00 0000 0000 0000 00 400353 1590 0355
06 იი ტილი იიიი იიიი 00 40 0355 1590 0357 00 00 000n onon nono 9. ტი 15.7 იიიი იიიი
00 41 1047 0000 1047 00 60 162 0344 0365 ი6 60 1941 ბიბი იიგი 94 60 noon in41 nnon
10 42 1039 1580 0000 00 40 1041 0000 1041 96 60 1586 0000 0000 00 47 1047 0000 1576
92 60 1047 0000 0000 08 60 0000 1047 0000 92 42 1039 1582 0000 ი 0 60 ბინი ბიიი იები
00 40 1047 0000 1047 00 40 1576 0000 1047 00 41 1563 1623 0000 00 63 0376 0247 0000
00 61 0000 0000 0000 00 60 1588 1571 0290 90 80 1041 0003 0000 no 80 1047 0002 0000
00 80 1576 0001 0000 00 80 1588 0004 0000 00 A 0000 0005 0000 იი თი იიიი იიი , იიბი
00 83 nooo 0007 0376 00 42 1036 1606 1568 00 42 114 1606 0000 00 70 1568 0000 0392
no 40 0385 1676 0385 00 40 1571 1621 1571 no 41 1564 Onnnnnnn იი 63 385 ივ 91 ივ w ]
no 60 1630 0385 0402 04 42 0385 1586 0000 00 40 1634 0000 0394 იტ ი იიიი გივი ჩიტი
00 40 0394 1590 0296 00 00 0000 0000 0000 no 40 0396 1590 0998 00nn onnonnon 0000
00 60 1630 0385 0000 0041 1047 1587 1047 00 60 1588 1571 0290 00 60 1681 0410 0404
00 60 1682 0410 0000 on 80 1576 0001 0000 on BO 1567 0002 0000 00 80 156 A onn3 0000
00 80 1588 0004 0000 00 80 1569 0005 0000 00 80 15.71 0006 0000 00 06 00 იი ლიიი 000 )
00 72 1587 1569 0153 96 60 1565 0000 0000 00 41 0124 0000 0124 ბი ბი ბლიი ბიბ იბი
00 41 1650 1623 0095 00 60 1664 0587 n000 00 60 1580 1563 0553 00 60 1667 0095 0000
00 60 1664 0547 0000 00 60 1563 1564 0526 on 60 167. 0453 0439 ბი 60 ბიიი ბიიი ბიბი
00 42 1036 1606 1568 00 42 1148 1606 1562 00 41 1568 0000 0000 00 63 0428 0000 0439
00 61 0000 0000 0000 on 40 24 1626 0424 00 40 1623 1571 1571 00 60 1148 1248 Onoo
00 40 0430 1629 0432 00 00 0000 0000 0000 no 40 0432 1629 0434 0000 0000 onoo 0000
00 40 0434 1629 0430 00 41 1564 1571 0000 00 63 0424 0454 0000 00 61 Onno 0000 0000
00 60 0000 0000 0000 on 42 1036 1593 1000 00 41 1036 onon 1036 0042 0430 1585 0000
00 40 1639 0000 0444 00 00 0000 0000 noon no 40 0444 1629 0446 იი იი იბი გიგი გიგი
00 40 0446 1629 0448 00 00 0000 0000 0000 04 60 1626 0000 0000 00 40 0430 0000 0430
00 60 1587 1574 0000 nn 40 1047 1587 1047 no 60 1640 0439 042A no 60 1614 0439 Onon
00 70 1587 1574 0457 00 60 1655 0453 0440 00 40 1564 1623 1564 00 60 1641 0430 0000
00 60 1588 1571 0000 on 42 0424 1582 0000 0 40 1630 0000 0424 00 60 1248 1148 0000
00 40 0462 1629 0462 00 40 1571 1587 1571 00 41 1564 0000 0000 00 63 0462 0468 0000
00 61 0000 0000 0000 00 60 1642 0462 0000 00 60 158 1571 0000 00 60 1588 1574 0290
00 12 1563 1588 0473 00 60 1679 0504 0489 00 42 1033 1606 1568 00 42 1048 1606 1562
0041 1568 0000 0000 00 63 0478 0000 0489 00 61 0000 0000 0000 00 40 04 74 1626 0474
00 40 16 23 1571 1571 00 60 1048 1248 0000 on 40 0480 1629 04 A2 ბი ბი ბეტი ტიიტ ბიჩი
00 40 0482 1629 0484 00 00 000 0000 0000 no 40 0484 1629 0480 00 41 1563 1571 იუბი
00 63 04 74 0505 0000 on 61 0000 0000 0000 00 60 0000 0000 0000 04 42 0480 15AS 0000
ია 40 109 ბიიი 0492 00 00 0000 0000 0000 96 42 0497 1582 0000 იი სი 1657 ბიიი ჩიტი
00 40 1590 0000 0496 00 00 0000 000 0000 00 40 0496 1629 049A 00 00 0000 Onno 0000
04 60 16 26 0000 0000 00 40 0480 0000 0480 00 60 1587 1574 0000 98 60 1587 ნიიი ბბბბ
06 40 1047 წილი 1947 no 60 1659 0489 04 78 00 60 1614 0489 0000 00 70 1587 15 74 050A
00 60 1661 004 0490 00 40 1563 1623 1563 00 60 158A 1571 0000 00 60 1662 0480 0000
0042 04 74 1582 0000 00 40 1632 0000 04 74 00 60 1246 1048 0000 00 40 0513 1629 0513
00 40 1571 1581 1571 00 41 1563 0000 0000 00 63 0519 0519 0000 იი 61 ბიიი მიიი იიბი
00 60 1673 0513 0000 00 60 1588 1571 0470 00 74 103 1042 0523 00 00 0000 0000 0095
00 42 0321 1585 0000 00 41 0521 000 0521 no 40 0521 1623 0521 00 42 1933 16n6 156 A
0042 1042 1606 1567 15 60 1033 1569 0000 00 72 1587 0000 0153 00 72 1568 1507 0545

UPDATING ROUTIK pago

00 60 1968 1042 0000 on 42 1036 1593 no no 00 41 1036 0000 1036 00 60 1036 114A 0000
00 40 0534 1629 0936 00 00 0000 0000 nono 0 40 0536 1629 0528 0000 0000 0000 0000
00 42 05 38 1505 0000 on 40 1590 0000 0000 00 40 1639 0000 0534 00 40 1564 1623 1564
00 40 1047 1587 1047 00 60 0000 0000 0057 00 60 1646 0573 0000 00 60 1649 0534 0000
00 60 1645 0224 0163 00 70 1570 1588 0619 00 40 1653 1587 0234 00 60 0000 0000 0225
00 74 0000 0000 0553 00 60 0000 0000 0095 00 42 1036 1606 1568 00 42 1042 1606 1567
15 60 1033 1569 0000 00 72 1587 0000 0153 00 72 156 1567 0585 00 42 1036 1593 0000
0041 1036 0000 1042 10 42 1039 1580 1041 00 60 158A 1043 0000 00 60 0000 1044 Onoo
00 60 0000 1045 0000 m 60 0000 1046 0000 96 60 1047 0000 0000 04 6 0000 1047 0000
88 42 1039 1582 0000 00 40 1047 0000 1047 07 42 1023 1643 0000 00 49 1047 Onno 1047
07 42 1033 1644 0000 on 40 1047 0000 1047 00 42 1033 1606 1048 96 42 0573 15 A2 0000
00 40 1590 0000 0000 00 40 1631 0000 0577 0000 0000 0000 0000 00 40 0577 1629 0579
00 00 000 0000 0000 00 40 0573 1623 0573 98 60 1587 onon 0000 00 4n 1047 0000 1047
00 40 1563 1623 1563 00 60 0000 0000 0057 00 60 1649 0534 0000 00 60 1646 0573 0000
00 60 1050 0224 0163 on 40 0198 1674 0198 no 41 0202 1675 0202 00 41 1563 1623 1563
00 41 0000 1577 0000 on 63 019. 0203 0000 00 61 000 000n nono no 41 0212 1674 0212
00 41 0216 1675 0216 00 41 1564 1623 1564 00 41 0000 1577 0000 00 63 0212 0217 0593
00 74 1676 0550 054 8 00 70 1572 1623 0616 00 60 1622 1572 0000 00 60 1637 on 57 0000
00 60 1614 16 38 0000 00 60 0161 0151 0000 00 72 1970 1588 0163 ბ) 60 6 იიი ბიბ ნი 96
C ) 70 1572 1622 0616 00 60 1623 1572 0000 00 60 1614 0552 0000 00 60 1672 0522 0000
00 60 1677 0234 0000 00 60 1033 1042 1000 00 60 1588 1563 0000 00 60 1588 1564 0000
00 60 1580 1041 0225 00 70 1570 1588 0618 00 60 0000 0000 0627 03 OA 0000 0000 0000
01 91 0000 00no onno 2 95 იიბი იიიი იიიი ი : 93 იიიი იიიიიიიი იიიიიიიი იიგი
00 60 0510 0001 0001 00 40 1663 1575 0625 00 00 nmon 0000 0000 იი ძი იიიი ლეი ) 102
00 74 1676 050 0549 00 60 0000 0000 0349 00 00 noon 0000 0000 no on onnn nono non
00 00 0000 0000 0000 00 00 onoo oonn 0000 იი იიიიიიიიიი იიიი 9 5 იიიი იიიი იიიი
39 09 Onno onno onnn 11 11 11 - ი იიიი იიიი იი ჩი იი 11 1111 იიიი on monn 0200 1111
00 01 000 noon onon 11 11 1111 11 - ი იიიი ბი აი იიიი 111 იიიი nnnn 1111 nnnnnnnn
00 00 0000 0000 0001 00 00 0000 0000 0000 00 60 0000 0000 0063 000 0000 0001 0000
00 00 0000 0000 0007 00 60 4629 1033 0000 5500 000n onoo non 66 00 0000 0000 0000
00 60 0000 0000 0102 იი იბ იბბბ ტლიბ იბ ) 1 12 00 0000 0000 0000 no 07 0000 0000 0000
Do 01 0000 0000 0000 00 60 6930 1041 0000 06 58 0001 0007 0122 06 58 0001 0000 0130
no 60 6830 1048 0000 n6 5A 9001 on 0138 no 60 6130 114A DOO იი იი ტი 11 111 116
00 42 1033 1606 156. 00 42 1036 1606 1568 00 00 0000 0000 0200 ბ) 6) იიიი ბიიი 0225
00 60 0000 0000 0172 ი იი ბიი 1 იიინ იცი იი 60 141 934 იიიი იი კი იიიი ტიიუ იგი
00 54 0000 0000 0000 0.58 0001 006 0196 00 60 104 A 9348 0000 OA 58 0001 0000 0212
00 60 1148 934 8 0000 01 91 13.0 138 იბი 1 08 58 0001 0007 0186 00 00 0000 0000 0002
0ტ იიიიიი იჩიი გივ იი იიიიიი ბიიი იბი იი ი ) იიბი იიიი ილი ; იი კი გიი 3 იიიიიიიი
00 42 1248 1606 0000 00 60 1037 0.000000 ჯი ბი ბიი ) იიბჯიბიი 00 42 114A 1406 0000
00 60 1034 0000 6000 bo 42 1048 1606 9000 00 60 0000 0000 0274 00 on 1593 Onnn 0000
00 60 0000 0000 0710 00 60 156 1577 0000 00 60 0000 0000 0096 ბ) ტი იბი ბიიი 145
00 60 1036 0000 0000 00 60 0000 0000 026 00 60 141 1240 ძიბი 00 60 1248 1148 0000
01 10 0000 0000 0000 იი 01 1600 ბიიი შიი 00 60 1588 1047 0531 do 42 1039 1606 1048
იი იმ ბილი ტბის ი 217 00 0 103 , 114 , გ 0 ტი
00 60 0000 0000 0203 00 60 158R 1047 0558
00 60 1045 0226 0000 00 60 1650 0224 0948 00 60 0000 0000 0618 00 60 1640 0439 042A

EXHIBIT VI

60
 UPDATING ROUTINE
page 5
00 60 1654 0453 0457 00 60 0000 0000 0471 no 60 1034 0000 0000 00 42 1033 16n6 noon
00 60 0000 0000 0478 00 60 1659 0489 0478 00 60 1660 0504 0508 00 60 1048 1248 0000
00 52 0000 0000 0000 00 60 1665 0224 054 8 00 60 0000 0000 0624 00 60 0000 0000 0418
00 60 0000 0000 0526 00 60 0000 0000 0384 04 58 0001 0007 0063 00 60 0000 0000 0133
00 60 noan Ono 0141 00 60 0000 0000 0526 no 60 124 A 104R 0000 00 00 0002 0000 0000
00 00 0000 0003 0000 00 60 1614 0224 0163 00 60 0000 0000 0070 00 60 1654 0453 0000
00 60 1660 0504 0000 00 70 1564 1588 0421 00 81 1564 0007 0401 00 81 1563 0007 0376
00 81 1564 0007 0271 00 81 1563 0007 0288 00 81 1564 0007 0308 00 81 1563 0007 0327
00 40 1702 0000 1694 92 60 0000 0000 0000 00 40 1703 0000 1690 00 00 0000 0000 0000
00 00 0000 Onno 0009 0000 0000 0000 000 on 00 Onon non n000 on on 000 0000 0000
00 61 0000 0000 1696 12 60 1692 1691 0000 88 42 1702 0000 0000 00 40 1701 0000 1690
00 60 1700 1693 1690 00 68 1695 1693 1691 88 40 1691 1702 1691 00 59 9999 9132 0001
00 59 9998 1692 1696 00 40 1723 0000 1716 91 60 0000 0000 0000 01 40 1724 0000 1708
00 60 1727 1718 0000 00 00 0000 0000 0000 00 68 0000 0000 0000 12 42 1721 1705 0000
88 40 1726 0000 1712 00 00 0000 0000 0000 00 41 1712 1725 1714 00 00 0000 0000 noon
on 6A 0000 Onno 1718 00 00 0000 0000 0000 00 61 1721 0000 1727 0000 0000 0000 0000
00 40 1721 1726 1720 00 00 0000 0000 0000 0000 0000 0000 0000 00 60 1717 1718 1709
00 59 9999 9140 0001 05 99 9981 7210 0000 00 01 0000 0000 0000 00 02 0000 0030 0097
00 60 1712 1718 1718 02 91 0051 1770 0001

* 944 Input - output locations inserted here .

EXHIBIT VI

and the entire last block of the event file is re fixed. Theoretically, there is no upper limit to the
tested . size of an event record . However , practical limita
In addition to the regular updating operations tion has been imposed by limiting a tape block
just discussed , the program performs several other length to 200 words , thus restricting the maximum
housekeeping functions . First , if transactions re event record length to one full block or 200 words .
main after the event file tapes are exhausted , each The static or fixed data in the file record require
remaining transaction is tested for a new activity 7 words . In addition , a single predecessor or suc
code (Code 1 ) and a successor updating code . If cessor reference requires 3 words , so that the
the test indicates a new activity and a successor minimum record length is 10 words . The maximum
updating, a new event is constructed and added to length of the record is 200 words , which limits the
the event file. Any other combinat ion of transaction event complexity to no more than a total of 64 prede
and updating codes will reject the transaction in the cessors and successors . This does not appear to
usual manner . be a serious restriction at this time .
As the run is being made, the program counts the The NORC is a fixed word machine . Thus , in
number of transactions processed into the event file
order to use the variable length record technique, it
After the run is completed , the program prints the was necessary to add three control words . The first
number of transactions successfully processed and
controls the length of the record itself . The re
the input and output tape labels at the end of the maining two control the number of predecessor and
reject register .
successor references within the record . :
The output tapes from the file updating run (Run
06 ) are first a new (updated ) event file tape , and A NORC word consists of 16 decimal digits
second , a completed events tape which lists the available for variable data usage and a 17th digit
complete file for each event that has had all its used as a word check digit . The check digit is
predecessor activities completed . used to verify tape and high speed memory transfers
The detailed program flow chart and absolute and is not available for actual use as data storage .
coding for Run 06 are to be found in Exhibits V and Although NORC is designed to handle only decimal
VI, on pages 57 through 61.
. digits , it is possible to generate and handle alpha
The outputs of Run 06 are of much importance to betic information with the NORC itself . However ,
the analysis . The completed events file represents two digits are required to represent each alphabetic
a detailed history of experience which will be avail character and certain programing restrictions are
able for separate analysis work. The event file required . Fortunately , alphabetic information is not
tape represents the current picture of future activity required in the PERT system .
It is the basis of the subsequent evaluation anal All data items in the PERT system are fixed in
ysis . The composition of the event file tape can length . The economy that might be achieved in file
now be discussed in detail . space by using variable length data items is far out
(4 ) Event File Tape . The information that is weighed by the large increase in programing com
kept in the event file is by nature variable in length plexity that would be incurred . There are very few
although it is made up of fixed length items . The data items for which the ratio of maximum length to
length of the file message or record is determined by average length is very large .
the number of predecessors and successors an event NORC will operate arithmetically either in fixed
has . In order for an event to exist in the file , there point or floating point rotation depending upon the
must be at least one successor or predecessor event actual programing codes selected . When operating
related to it. The minimum size of a record is thus in the floating point mode , a full word is required

61
 DATA ITEV REFERENCE
FILE LAYOUT CROSS- REFERENCE

Number Word Word

Data Item of Digits Number Position

1. Message length 3 1 1 - 3

2. Probability P_{T, > T ,> Tg)

( ΣΤg 2 1 5 - 6

3. Sign T - T.
TE 1 8 2

4. 4 1 7 - 10
TL - TE
5. Schedule date 6 1 11 – 16

6. Predecessor counter 2 2 1 - 2
7. Successor counter 2 2 3 - 4

8. Probability P ,(TTS)
Er
2 2 5 6

9. Event code 9 2 7- 15

10 . Report code 1 2 16

11 . IE 16 3 1 - 16

12. o?0²т 16 4 1 - 16
5B E

13 . TL 16 5 1 - 16

14 . 0²7 16 6 1 - 16

15 . Sign Ts 1 1 4

16 . Ts 4 7 1- 4

17 . Completion week 4 7 5 – 8

18 . Resource code 2 7 9 – 10

19 . Specification code 2 7 11 - 12

20. ' Predecessor index 2 7 13 – 14

21 . Successor index 2 7. 15 - 16

22. Predecessor code 9 8 7 – 15

23 . te ( Predecessor ) 16 9 1 - 16

o 16 10 1 - 16
24. O. (Predecessor)
25 . t . ( Successor) 16 11 1 - 16

o 16 12 1- 6
26. ? (Successor)

EXHIBIT VII

62
 EVENT FILE TAPE LAYOUT

WORD .
# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

MESSAGE
LENGTH Ts P TL - TE SCHEDULE DATE

>
X

X
x x х х х х х x x • х х Х - X X - X х х
PRED . SUC . REPT
COUNTER COUNTER P, EVENT CODE CODE ✓
2
x x x x х х x x x x x x x x x х

*
o
TE ?
3

> *
х x x x x x x x X X X X
X X 1 X 1 X
1 OTE

х
X x 1 x 1 x x x x x x x x x x x x

< xR
X

. TL

i
5
х X
1 1 x х х х х х х x x x x
6
1 1 1,2
х x 1 x 1 x х x x х x x х х х х х х х
COMPLETION RESOURCE SPEC .
PRED . IN .
Ts DATE CODE CODE CODE SUC . IN.
7

х
х X х х х x x х х .
х х Х х х х х

+ PREDECESSOR CODE
8

>
х

х o O O X x x x х х х х

O
O
o

*
te

х
X x 1 x 1 x x x x х x x x x х х Х х

O ✓
10
x 1 x 1 x х х х X x x х х X X х Х х
X

<x
11
X

х х Х х х x x x x х х x x
x | x | x X

✓
12
х x 1 x 1 х х х x x х х х X X X х

1 ✓
.

13
X
X

X X | 1 x х х х х х х X х х x
X
to

14
1 ✓
X

х x | x | x х х х х х х х х Х х х х х

EXHIBIT VIII

for each data item . The first two digits of the word Two restrictions were made on the actual layout
specify the location of the decimal point . The third of the event file . First , it should be as compact as
digit indicates the algebraic sign of the number and possible , consistent with good coding techniques ,
the remaining 13 digits , the significant digits of the and second , no data item should be spread over two
number . In order to preserve as much precision as words . The event file tape is summarized in Exhibit
possible throughout the PERT calculations , it is VII, on opposite page, and shown in layout form in
necessary to perform some of the arithmetic opera Exhibit VIII above .
tions in the floating point mode . Data items refer The basic event file tape has now been gener
2 2
ring to Te TL ote: 0 Lite, and o are all ated . The next step in the cycle is the analysis
carried as fuit NORc words in the floåting point run .
mode . The remaining numeric data items are carried
in a compact fixed point notation .

63
 IV . THE PERT COMPUTER ANALYSIS

1. THE COMPUTING RUN the next successor of the first event on Tape 2 be
( 1 ) Sequencing (Run 07) . The method of elapsed comes the search criterion ; if the comparison is
time computation as outlined in the Phase I PERT equal the first successor of the next event in Tape
project report outlines in detail the requirements for 2 becomes the search criterion .
the sequence of events needed to perform the calcu The search on the input tape ( Tape 1 ) continues
lations . It requires that calculations for the events with transfers to the intermediate work tape ( Tape 5
proceed in accordance with an ordered sequence . or 6 ) . When searching Tape 1 , if the search com
This ordered sequence essentially collapses the parison becomes negative (Tape 2 minus Tape 1 )
PERT network into a single line . The line of then the remainder of the block is transferred to the
events is constituted so that all arrows point in the intermediate work tape ( Tape 5 or 6 ) . The work
same direction . Thus , it is necessary to arrange tape just written on is rewound , and the search
the events within the machine in such a particular begins on that tape . As either Tape 5 or Tape 6 is 1
sequence . The computing sequence run accom searched , the tape being searched is written on the
plishes this by mechanizing the method outlined in other work tape . For example , if Tape 5 is being
the Phase I report . In order to mechanize this searched , the information is also transferred to
method , a few changes in details were necessary. Tape 6. As an item is found on a work tape and all
Essentially , however , the method used is the one the predecessors are satisfied ( i.e. , the predeces
described in the Phase I report . sor counter equal to the predecessor index ) , they
The only input required is a file of events , ar are transferred to Tape 4 or Tape 2. Tapes 2 and 4
ranged in any order . ( Most frequently they will be are used alternatively as the storage for events
arranged in event code order , since the input to whose predecessors are satisfied . If Tape 2 is
Run 07 will generally be the output of Run 06. This being read , then Tape 4 is used for storage . If
is not a necessary condition , however . ) Tape 4 is being read , then Tape 2 is being used for
The first operation of the program is to scan the storage .
input tape ( Tape 1 ) for events having no predeces If the search criterion ( Tape 2-Tape 5 or 6)
sors ( the initial events in the sequence ) . These gives a negative result with one of the work tapes ,
events are extracted from the input tape and put on the search will stop . The remainder of the file on
the tape to be used for events for which the above the tape being searched is then transferred to the
predecessor condition has been satisfied ( Tape 2) . other work tape . The search will then begin on the
Both tapes are then rewound . The first event on ot her work tape after it has been rewound . If the
Tape 2 is extracted , as well as its successor code . end of a work tape is reached and there has been a
The input tape ( Tape 1 ) is searched for the event previous negative search test on the other work
which matches the successor of the first event on tape , the search will then begin on the input tape
Tape 2. The matching operation is performed ( Tape 1 ) . Note that Tape 1 is never rewound . The
through a subtraction . The event from the satisfied search on the input tape always progresses from the
predecessor tape , Tape 2 , is subtracted from the last stopping point .
event on the tape being searched . Thus , as long as Generally , events flow selectively from the input
the result of the subtraction remains positive , the tape ( Tape 1 ) , to the work tapes (Tapes 5 and 6 ) .
search continues . A zero result indicates the re From the work tapes , the events flow to the tape
quired event has been located . A negative result which contains all events for which the predeces
indicates that either the network was constructed sors have been satisfied ( Tapes 2 and 4 ) . When all
illogically and is in error or that the program must the successors of the events on Tapes 2 and 4 have
branch and search elsewhere for the required event . been satisfied , the events flow to the output tape
On the first search on the input tape ( Tape 1 ) , ( Tape 3 ) and all the sequencing criteria necessary
the event must be found or the negative result from for computation are satisfied . Tape 3 is the input
the searching test indicates an error . During the for the succeeding run ( Run 08 ) .
search for the first event on the input tape , the
A standard stop for the routine is reached when
input tape , Tape 1 , is transferred directly to one of
the intermediate work tapes ( Tape 5 or 6 ) . When both Tapes 2 and 4 are empty and the input tape
the event is found on Tape 1 , its predecessor is ( Tape 1 ) is exhausted .
compared to the search criterion . If they are equal , In relating the program to the method described
the predecessor counter is increased by one and in the Phase I report , it can be noted that Tape 1
compared to the predecessor index . If the latter are and Tapes 5 and 6 are equivalent to Pile 1. Tape 2
equal , the event is transferred to the conjugate or
continuation of Tape 2 , Tape 4. If they are unequal , and Tape 4 are equivalent to Pile 2 ; Tape 3 is
the event is transferred to the intermediate or work equivalent to Pile 3. Exhibit IX , on the following
tape ( Tape 5 or Tape 6 ) . The successo: counter of page , gives a simplified flow diagram of the program .
the event from Tape 2 , the search criterion , is in
A detailed program flow chart and absolute coding
creased by one and compared to the successor
index . If these two are equal , the event is removed for Run 07 is found in Exhibits X and XI , following .
from Tape 2 and transferred to the final sequenced In addition to listing the beginning and ending
output tape , Tape 3. If the comparison is unequal, tape labels , the computing sequence run (Run 07)

64
 INFORMATION FLOW DIAGRAM
COMPUTATION SEQUENCE RUN (RUN 07 )

TAPE TAPE
TAPE 5 6

11
+
[ ] PILE

o
SEARCH
CRITERIA

TAPE
TAPE
PILE
-

[ ]
-
|

TAPE

PILE

[
EXAINT IX

65
produces a total of the number of events on the for each event . The more complex events require
final output tape . 20 to 30 words of tape and memory storage . For
( 2 ) Slack and Probability Computations (Run 08 ) . computation purposes , the working tape using 3
Once a sequenced event file has been prepared, the words per event is equivalent to the full event file .
computing of the outputs can follow immediately . After computing the Tg and ožeE for the first and
The logic of the computing run is completely de following “ no predecessor" events , the program
pendent upon sets of sequenced events as input . examines the next event , the first event with prede
During the forward analysis of the sequenced cessors . The program selects the first predecessor
event file , Te and ože are calculated and stored in
E
and goes back to this predecessor event and tests
the file . After the Te and of g have been calcu
E E for the presence of completion date . If there is a
lated for the last event in the file , the process is completion date , the program picks up the report
reversed . The sequenced net of events is analyzed week constant and computes the difference . This
in reverse by reading the tape in reverse . Now TL difference is then applied to the predecessor t e as
and oš , are calculated and stored on the event file . a correction for time “ now . " The proportional
As the Tį and o* , calculations are completed for change in t , is applied to o
e to adjust the vari
each event , the program branches and calculates ance . With these adjustments made , the computa
the slack TL - Te . The value obtained for the tion of Te and ož E is extracted . The sums are
slack TL - Te is tested and if the result lies be computed and the results stored . The process is
tween 0 and 5 , it is flagged by putting a 9 code in repeated for each of the predecessors of the event
the report code field . This will allow the event to being considered . The stored results for the nomi
be selected automatically for special output re E s are comp
nated Te's and ofe' ared . The larger
porting . Te is selected and it , along with its oil , are
L

In order to operate , the program needs only one stored in the main event file . The temporary work
input tape , a file of sequenced events , and the ing tape message is constructed and read out . The
output of the sequencing run ( Run 07 ) . The program next event on the event file is extracted and the
then tests for the presence of a scheduled time Ts . computation cycle is repeated .
If a Ts exists in the file for the event, the proba The adjustment subroutine , completed or begin
bility calculation for P ,( Ts < TE . ) is made . In ning events , is only repeated when reference is
addition , the calculations P (TL .' > Ts ) and made to beginning or completed events . If the event
P ,( TE : ETs < TL . ). ( Te : and TL are random
! 1 is a beginning event and no completion date is
variables . ) carried , the time adjustment is made from time zero .
The first block on the input tape is read into The time adjustment is so programed as to prevent
memory and the first event is extracted for computa the adjustment on a te value to less than one week .
tion . The first event on the tape by definition will The forward computation cycle continues until
have no predecessors , this also could be the case the input file is exhausted . The last event on the
for one or more of the events follow ing the first file is assumed to be the objective event and all
event in the file . The program goes through a event files and networks must be adjusted to reflect
dummy calculation cycle and develops a zero TE
2
this condition . After computing Tg and ote for the
E E
and OÍ E The program then constructs a working last or objective event , the program tests two option
tape for later use in the calculation. Three words switches. If the first switch is “ on , " the program
are read on to this working tape , the event code , the will select the Ts for this objective event and set it
Te value , and the of E value . The nine digit event up as the Tl for the last event . If the first switch
code is read into the first word of the working tape is “ off , " the program will test the second switch . If
message. Te , in full NORC word floating point this switch is “ on , ” the program will pick up a
form , is read into the second word of the working manually inserted value and set it up as the Tl for
tape message and of E ' again in full NORC word the last event . If both switches are off , the pro
floating point form , is read into the third word of gram will select the Te for the last event and set it
this working tape message . up as the Tl for the last event .
The working tape is a method used to decrease The working tape used in the forward calculation
the size of the file needed for computational look- up . has no further value to the operation . However , a
As such , it increases the efficiency of the program new working tape is set up using the event code TL
and reduces the overall running time substantially . and 02TL in the same manner as before . Before the
The full event file contains as a minimum 10 words program goes into the next event in the reverse

66
 PROGRAM FLOW CHART
SEQUENCING RUN #07

CLEAR MOD . STOP

READ IN TPE LABEL COUNTER SET

(01 ) CO.AP À E LABEL RESET 42
WITH KEYED IN DATE
SENSE = 0

0 0

STOP PRINT
STOP 100

301
READ INPUT FRO
TAPE
DUNTER
# 11 RESET

EOF

NO 121
EXIT 344

122
EOB - STORE, COMP .
BLOCK LENGTH =
GET EVENT INDEX = Σο
EXTRACT PRED . INDEX
SENSE = 0 190

SENSE SUC . INDEX = 0 SENSE READING 01, 05

OR 06 TAPE
05 OR 06
= 0 0
SET E
01

65 SET EXIT # 2

COPY BLOCK IN STORAGE

ON TO OS OR 06 TAPE
SET EXIT # 121

EXIT

300
R EVENTS SENSE EOF ON 05 OR
FLAGGED 06
PE . SET
STORE EVENTS W1
PRED . INDEX = YES
OUTPUT STORAGE
200 WORDS SET EXIT 276
465

0
CHANGE OUTPUT TAPE TO
INPUT AND INPUT TO
OUTPUT OS OR 06
WRITE EOF ; REWIND
WRI 1 450
VER

344

READ OS OR 06 TAPE
# 5
TCF

NO
EOF
EOF ON 02 REMND
01 AND 02 CL EAR MI
SET EXIT # 11 YES

SENSE ODEOF COUNTER

100
195 3

SET EXIT 122 REWIND

READ TAPE STOP

301

EXHIBIT X
 SEQUENCING ROUTINE page 1
-02 91 0051 0056 0000 09 96 0008 1997 0001 00 6A 0000 0000 0006 00 60 0000 0000 0008
00 00 0000 0000 0000 00 61 0000 0000 0001 00 on noon 0000 3953 02 01 0051 0056 0000

02 91 0051 2040 0001 03 94 0650 0650 0000 08 6A 0009 onoo 1931 00 61 Onno 0000 0615
02 91 0000 0000 0000 08 60 0012 0000 1948 00 67 0000 0000 0015 00 61 0000 0000 0000
00 57 0000 0000 0000 00 40 1729 1687 1687 00 70 0650 0000 0019 00 61 Ono 0000 000
00 0 1748 729 1729 00 40 1686 0000 1686 01 91 1686 1686 0000 08 60 0022 00 no 1948
03 94 0684 0000 1000 08 68 0024 0000 1931 00 67 0000 0000 0265 00 60 0000 1689 0000
04 42 1689 1730 1690 08 42 1689 1731 0000 00 41 1690 0000 1691 00 40 1732 0000 1691
00 51 0000 0000 0000 00 40 0023 1732 0023 00 41 1691 1692 0000 00 63 0039 0136 0000
00 61 0000 0000 0000 00 60 1733 1692 0000 00 60 1743 0052 0000 00 60 0000 0000 0023
13 60 4684 1696 0000 02 42 4690 1734 1694 00 42 4690 1735 1695 00 70 1694 1733 0044
00 60 1770 0079 0045 on 60 1 744 0079 noon 96 34 1694 1747 1694 96 34 1695 1747 1695
on 42 1694 1746 1694 00 42 1695 1746 1695 on 40 1692 1696 1692 00 60 4684 0A85 0000
00 40 0050 1737 0050 04 58 0001 0007 0050 00 60 1738 0050 0000 00 72 1694 1733 0056
00 60 0000 0000 0064 96 60 1694 0062 0000 00 42 0052 173 0000 00 40 0062 Onn 0062
00 40 1739 0000 0062 00 60 4684 0892 0000 00 40 0060 1737 0060 00 00 0000 0000 noon
00 60 1740 0060 0000 00 72 1695 1733 0066 00 60 0000 0000 0077 00 59 Onno 0000 noon
04 42 0000 1731 0073 96 60 1695 0000 0000 00 40 0073 0000 0073 00 40 1741 0000 0073
00 60 4684 0988 0000 00 40 0071 1737 0071 00 00 0000 0000 0000 00 60 1742 0071 0000
00 60 1733 0887 0000 00 60 1733 0888 0000 96 60 1096 0000 0000 00 40 0052 00no 0052
00 00 0000 0000 0000 08 42 0891 1731 0000 00 40 1745 0000 1697 00 42 1686 1746 0000
00 40 1745 0000 1698 50 22 0000 1697 1699 00 60 0989 1700 0000 00 70 1700 1733 0097
50 22 1700 1699 1701 50 22 0000 1748 0000 00 63 0092 0092 0000 00 60 1748 1701 0092
00 00 0000 0000 0000 50 26 1701 1700 1702 00 60 1701 0989 0000 00 60 0990 1700 0000
50 24 1702 0000 1701 00 60 0000 0990 0000 00 40 0085 1747 OOA 5 00 40 0093 1749 0099
00 40 009 1747 0094 00 40 0096 1749 0096 04 60 1749 0000 0000 00 40 1703 0000 1703
00 41 1695 0000 0000 on 63 0085 0106 0000 00 61 0000 Onon 0000 00 60 1733 1703 000n
00 60 1750 0085 0000 on 60 1751 0093 0000 00 60 1752 0094 0000 00 60 1753 0096 0000
00 60 1733 1700 0112 00 54 0000 0000 0000 00 60 1764 6652 0000 00 40 1704 1732 1704
00 60 8885 6653 0000 00 50 0000 0001 0000 08 58 0001 0007 0115 00 54 0000 0000 0000
96 60 1694 0000 0000 00 40 1756 0000 0124 00 70 1694 1733 0125 00 60 A892 6653 0000
00 50 0000 0001 0000 00 00 0000 0000 0000 00 54 0000 0000 0000 96 60 1695 0000 0000
00 40 1758 0000 0132 00 72 1695 1733 0130 00 60 0000 0000 0133 00 60 8988 6653 0000
00 50 0000 0001 0000 00 00 0000 0000 0000 00 50 0000 0001 0000 00 40 1696 1704 1704
00 60 0000 0000 0146 96 60 1704 0000 0000 00 41 0000 1737 0000 00 40 1760 0000 0139
00 00 0000 0000 0000 08 60 0140 0000 1948 00 40 1760 1732 1760 00 60 1733 1704 0000
00 52 0000 0000 0000 00 60 0000 0000 0036 00 60 0000 0000 0000 00 40 1705 1749 1705
on 41 1763 1705 0000 on 63 0156 0155 1152 00 60 1732 1705 0000 00 40 1768 1767 016A
00 60 1766 0157 0153 00 61 0000 0000 0000 00 60 1836 0164 0000 00 60 1767 0166 0033
00 60 1769 0170 0157 00 60 1765 0170 0000 00 42 0886 1764 1084 04 60 1749 Ono 0000
00 40 0157 0000 0157 11 34 1699 1748 1699 89 42 0000 1746 1726 98 42 1699 1825 0000
00 40 1726 0000 0000 00 40 1084 0000 1084 00 40 0164 0566 0164 00 60 0887 1085 0000
00 40 0166 1749 0166 00 60 0888 1086 0000 00 40 016A 1749 0168 00 00 0000 0000 0000

SEQUENCING ROUTINE page 2

04 91 1084 1683 0001 08 60 0172 0000 1948 00 40 0171 1732 0171 00 00 0000 0000 0000
96 60 16 94 0000 0000 00 40 1772 0000 0181 00 54 0000 0000 0000 00 42 1084 1764 1706
0042 8892 1764 1707 00 70 1706 1707 0187 00 00 0000 000 0000 00 40 0178 1747 017A
00 40 1700 1749 1700 00 41 1763 00 0000 63 01 75 0241 0000 no 61 0000 0000 0000
04 42 01 78 1731 0000 00 40 1737 0000 1699 00 59 0000 0000 0000 92 42 0000 1746 0000
00 40 1773 0000 0000 00 40 1699 0000 0218 00 42 0170 1731 1726 00 40 1837 0000 0195
00 00 0000 0000 0000 00 70 1728 1733 0218 00 40 1745 0000 1728 00 40 1726 1841 0199
on 00 0000 0000 0000 00 40 1728 0000 1728 00 42 021A 1731 0091 00 40 1842 nonn 0203
00 00 0000 0000 0000 00 40 0091 1843 0205 00 00 0000 0000 0000 50 22 0000 1748 0000
00 63 0209 0209 0000 00 60 1748 1701 0000 04 60 0091 0091 0000 00 40 1844 0091 0211
00 00 0000 0000 0000 00 40 1845 0091 0213 0000 0000 0000 0000 00 40 1846 0091 0216
00 40 1847 0091 0217 00 00 0000 0000 0000 00 00 0000 Onon 0000 0000 0000 0000 000
00 40 0218 1768 0000 00 40 1732 0000 0221 00 00 0000 0000 0000 50 22 1701 1708 0000
00 63 0000 0226 0228 00 60 1701 1708 0000 00 60 1702 1709 0228 50 22 1702 1709 0000
00 63 0224 0000 0000 00 40 1749 1703 1703 04 60 1703 0000 0000 00 41 1694 0000 noon
00 63 0182 0233 0000 00 61 0000 0000 0000 00 60 1732 1703 0000 00 60 1774 017A 0000
00 60 1733 1700 0000 00 60 1708 08 87 0000 00 60 1709 0888 0000 00 60 1733 1708 0000
00 60 0000 1709 0000 00 60 0000 0000 0112 00 74 0000 0000 0000 on 47 0171 1746 1710
00 40 0254 1710 0260 00 41 1710 1732 0000 00 40 0249 0000 0252 00 60 1775 01 74 0000
00 60 1794 0241 0171 04 91 0000 0000 0000 04 95 0000 0000 0000 00 67 onoo 0000 0252
00 61 0000 0000 0000 on 00 0000 0000 0000 00 67 0000 0000 0251 04 94 1084 0000 0000
08 68 0255 0000 1931 00 67 0000 0000 0251 00 41 0252 1732 0252 00 60 1795 0240 0000
00 60 1796 0241 0175 00 00 0000 0000 0000 08 68 0261 0000 1931 00 67 0000 0000 0251
00 60 1744 0241 0000 00 60 1797 0240 0112 00 40 1777 1777 0000 00 40 1767 0000 1767
00 40 1767 1768 0168 00 60 1807 1765 0000 00 60 1808 0258 0000 00 60 1767 0166 0000
00 60 1757 0154 0000 00 60 1733 1705 0000 00 60 1848 0160 0000 00 60 1744 0153 0000
00 60 1809 0259 0000 00 60 1810 0264 0000 00 60 1778 0157 0000 00 42 0139 1746 0000
00 40 0287 0000 0287 02 91 0000 0000 0000 02 95 0000 0000 0000 00 67 0000 0000 0284
00 61 0000 0000 0000 03 98 0000 0000 0000 04 98 0000 onon 0000 00 57 0000 0000 0000
02 95 0884 0000 0000 08 68 028A 0000 1931 00 67 0000 0000 0622 00 6n onno 16A9 Onon
04 42 0000 1730 1690 00 41 0287 1732 0287 08 42 1689 1731 0000 00 41 1690 0000 1691
00 40 1691 1732 1691 00 60 0000 0000 0000 00 41 1691 1692 0000 00 63 0303 0444 0000
00 61 0000 0000 0000 00 60 1733 1692 0000 00 60 1781 0306 0000 00 60 1771 0304 0287
00 54 0000 0000 0000 96 60 1691 0000 0000 00 40 1776 0000 030 A 00 70 OAA4 1764 0310
00 41 0306 1777 0306 00 00 0000 0000 0000 00 61 0000 0000 0000 00 42 0 306 1731 0000
00 40 0000 1777 1711 00 440306 1777 0306 00 60 1761 0304 0000 00 60 0000 Onno 0000
00 40 1711 1779 0000 00 40 1780 0000 0317 0000 0000 0000 0000 02 42 1693 1734 1694
00 42 1693 1735 1695 96 34 1694 1747 0000 00 42 0000 1746 1694 96 34 1695 1747 0000
00 42 0000 1746 1695 00 60 1736 1696 0000 00 40 1694 1695 0000 00 40 1696 0000 1696
00 40 1692 0000 1692 00 40 1732 1692 1692 96 40 1696 1732 0000 00 41 0308 0000 0308
00 51 0000 0000 0000 on 40 1782 1711 0333 00 60 4000 08 85 0000 00 40 0333 1737 0333
04 58 0001 0007 0333 00 42 0333 1731 0000 00 40 1783 0000 0343 96 60 1694 0345 0000
00 42 0335 1730 0000 00 40 0345 0000 0000 00 40 1784 0000 0 345 00 70 1694 1733 0346
00 00 0000 0000 0000 00 40 0343 1737 0343 00 00 0000 0000 0000 00 42 0343 1731 0000
0040 1786 0000 0353 96 60 1695 0355 0000 00 42 0345 1730 0000 00 40 0355 0000 0000

EXHIBIT XI

69
 SEQUENCING ROUTINE page 3

00 40 1787 0000 0355 on 70 1695 1733 0356 0000 0000 Onon anno on 40 0753 1737 0357
non non nono 0000 on 72 1695 1733 0453 on 76 noon onon 360 on 76 anno nnnnn364
on 60 anoo noon 0367 on 60 000 annnnnn 12 60 991 000 000 ‫ܘܪ‬ ‫ ܟܘܘܪ ܟܟܐܙ ܘܝ‬171
on 60 1788 0360036R on 60 000 onnn Onn ‫ ܘܪ‬66 168 ‫ܐ ܟ‬7‫ܘܘܘܘ ܕܠܐ‬ 00 en 17AA0364 034
0 60 0887 1712 0655 on 6 anno Onnnnnnn ‫ ܪܘ‬6‫ ܙ‬71‫ ܘ ܕܐ‬RAs non ‫ܪܪ‬ ‫ܙ ܘܪ‬337 ݂‫ ܘ ܘܘܕ݂ܶܟ‬non
nn on 1789 0367 0373 on 60 1 712 0889 non 2 60 1731 0000 0000 on 42 ORAS Onno non
‫ܘܟ‬ ‫ ܀‬1 ‫ܘ‬988 pnnn 1885 50 22 0889 08 A7 1697 50 29 noon 1733 000 50 72 1790 Onno annn
00 63 0384 0000 000 50 22 1697 1733 Ono 00 63 0000 0000 0632 50 22 1685 1697 0000
00 63 noon ono 0632 ‫ܙܩ‬ 6 ‫ܘ ܘ‬888 06‫ ܟ ܘܪ‬nno 99 66 Annn 18 86 ‫ܘܘܘܘ‬ ‫ܪܘ‬ ‫ ܘ ܘܝܐ‬B86 ‫ ܙ‬7‫ ܙܟ‬18 88
00 60 0000 0000 0389 on 000000 nnnnnnnn on 42 1697 1825 nono on 7 ? TA 25 0000 7393
50 30 1697 1833 1697 0 60 1748 1726 0395 on 60 1733 1726 non an 30 1697 1790 1697
00 34 1792 1697 1697 94 60 0000 0000 0000 cn ‫ܘ ܘ‬8 A ‫ ܘ ܘܘܘܘܟ‬83‫ܟ‬ 12 60 0891 ‫ܘܘܘܘ ܘܘܘܘ‬
00 72 1733 0000 0497 on 60 0000 Onno 1471 no 54 000 000 000 92 ‫ ܀‬1 61‫ ܟܘ‬17 1 ‫ܟܟܘܟ ܕ‬
60 ‫ ܘ ܟ‬0468_6066_1793 00 72 1695 1733 n4n6 00 60 0000 0000 0411 9660 1695 000 0000
46 ‫ܐ‬897 ‫ܘ ܘ ܘܘܘܘ‬1‫ܘ‬ no nonnon non n000 ‫ܝ ܘܟ‬1 ‫ܐ ܘܘ ܝܐ‬87 ‫ܘܘ‬R 00 onnnnnnnnnnn
no 54 0000 nnn 0000 Q‫ܕ‬ ‫ܐ ܙ ܀‬444 ‫ܕ‬ ‫ܘܘܘܘ‬ ‫ܕܢ ܟܟ‬ 93‫ܟܕ ܝܘ ܪܟܟܟ ܘ‬ ‫ܟܟܟ ܕ ܟ ܟܟ‬8 ‫ܕܐ‬1‫ܟܟܟܟ ܘ‬
no 40 042 000 0420 nn 7‫ ܕ‬61‫ ܙܐܕ‬133 14 1 8 on an noonnon 0427 96 1694 Onno non
on 40 180000 0422 ‫ܘܪ‬ ‫ܪܟ ܘܙ‬0‫ܟܪܘܘ ܘ‬ ‫ܟܟܟܟ‬ on 41 047 176R 0420 nn anno anno non
0 54 000 annon0 ‫ ܙ ܟ ܕܝ ܟ ܕ ܢ ܘܘ‬71c ‫ ܪܘ‬qn ‫ ܙ ܘ ܝ ܘܟ‬4‫ܙܪ‬ non n426 ‫ܟܘ‬ ‫ܟܟ‬ ‫ܟܘܘܪ‬ ‫ܟܟ ܟܟ‬ ‫ܟܟܟܟ‬
00 41 0426 1768 0426 NA 58 0001 0007 0426 ‫ ܪ܀ ܟܘ‬172 ; 173 ? 175 ? on 42 0429 1730 non
96 40 1759 0000 0439 00 40 JA 34 0000 0438 A 420on 1731 D000 00 40 047A ono 043A
no 40 94 39 1835 0439 470000 1731 nono 00 40 0430 00 0439 nn ‫ܘ‬ ‫ܟܟܟ ܙܟܟܘ‬ ‫ܟܘܟܘ‬
no no_onn anno qnon 00 40 1696 1704 1704 7660 1696 Onon 0000 ‫ ܀ ܟ‬1 179 nnnn_3 1793
on 60 0000 0000 0146 92 41 1704 1737 000 ‫ ܐ | ܝ ܟܟ‬802 ‫ܘܘ‬ ‫ ܟ ܘ‬6‫܀‬ ‫ܟܪܪܘ ܟܟܘ ܟܟܫܘ ܪܘ ܟܘ‬
OR 60 0447 0010 1948 00 40 802 1732 1802 96 60 1704 0000 0000 00 40 1793 nono 1797
00 60 1737 1704 0000 00 60 0000 00nn nico 96 60 1605 0000 0000 00 40 183 0000 0450
on 52 000 anno Ono ‫ ܟܘ‬2‫ ܀‬1‫ܘ‬8‫ ܟ‬1764 ‫ ܐܐ‬nt no 42 629A 1764 1707 ‫ܟ‬ ‫ ܐܐ ܟܐ‬n 6 ‫ܛܘ ܟܟܪܟ‬66
10 an nonn anno anno 00 40 0456 1747 74 56 no 40 1700 1740 1700 ‫ܝ ܟܘ‬1‫ܐ‬361 ‫ ܘܘܘܘ‬non
no 63 0453 0241 0000 no 61 noon On Onn 00 42 0456 1731 Onn ‫ܟܘ‬ n 1777 Annn ‫ܘܘܟܐ‬
no 59 noon nonnnnnn ‫ܐܟ ܟܟ‬ 171 ‫ܟ‬ 1 ‫ ܐ‬11 ‫ ܙ ܟ ܝ ܘܟ‬8 ‫ܟܟܟܘ ܟܟܟܟ ܪܘ‬ on 4 1609 nnnn n471
on nonnon 0000 Onon ‫ ܘ ܟ ܟܘ‬111‫ ܊‬18 ‫ܟܟ‬ ‫ܟܕ ܟܟ‬ in ‫ ܟ ??? ܙ ܟܟ ܟܙ ܟܟ‬non on 40 74 75 annn n4 75
‫ܘܘ‬ 00 noon so on Goon 00 60 1 744 0476 04 79 ‫ ܕܕ ܘܟ‬170 ‫ ܕ‬1798 ‫ ܘܘ‬on 00 63 04A3 04 Al 1470
on 60 1701 1708 0000 6 6 ‫ ܘ‬671‫ܕ‬71‫ܕܟ‬ ‫ ܝܐܕ‬83 ‫ ܕܕ ܘܟ‬71‫ ܕ ܕܟ‬7‫ܘܘܟܘ ܘܙ‬ 65 6 1 ‫ܟܘܟܘ ܪܟܟܟ ܘܐܟܪ‬
60 ‫ ܕܪܐܠ ܘܝ‬1113 ! 763 04 60 1703 non non no 41 1695 000 0000 on 63 044 04AA0non
00 61 0000 0000 0000 06 6 6 ; 33 1763_0000 00 60 1774 0456 0000 00 60 1733 1700 0000
00 60 1708 1889 onoo 0 60 709 0690 0000 on 60 1733 170A 0000 non anno 1709 noon
00 60 1806 0475 0000 on 60 0000 0000 0373 12 66_984‫ ܙ‬71‫ܕ‬non_3 ‫ ܝ ܘܘ‬2 ‫ ܘ‬R 3‫ܐ ܟ‬7‫ ܝ‬A non
00 40 1713 0000 1713 00 401745 0000 1713 50 270000 1684 1713 9660 OR 1 0200 noon
‫ ܀ ܟ‬66 ‫ܘܪܘܘ‬ 8‫ܘܟܟܘ ܙ ܘ‬ 0042 085 174 8 nnnn ‫ ܘܟ‬68_21 85 ‫ ܘ ܟܘܟܟ‬A A 5 ‫ ? ܝ ܘܘ‬1 713 1 825 ‫ܘܘܘܘ‬
00 72 0000 1825 0511 no 40 0885 174 8 0885 50 30 1713 1837 16A7 on on onno ono 0512
50 30 1113 a ?? ‫ ܘ‬168 7 00 34 1792 16 A7 1687 A8 60 1687 Onno 0000 no 40 A91 Onno 0891
00 60 1744 0561 anon on 72 89 1733 0518 on 60 174 A 1714 0520 on 600519 Anno 1972
‫ܐܙ‬ 1 ‫ܪܟ‬in ann 986‫ܟ‬ ‫ܟ‬ ‫ ܐ ܕܐ‬713 ‫ܘ‬8 89 161‫ܕ‬ ‫ ܘܟ‬26 1666 1714 1715 50 2A 1A12 1715 nnon
on 63 05 26 ono non ‫ ܟܟ‬2‫ ܀‬1 71 ‫ܟ‬ 1 826 anna 16 ‫ ܟ ܝ‬1 8 13 1‫ܪ‬0‫ ܟ‬171 ‫ܟ‬ ‫ܘܟ‬ 2‫ ܀‬111 ; j A ‫ܟܕ‬ ‫ܟܘܟ‬
on 72 noon 1825 0530 ‫ ܟ ܟ‬7 ‫ ܙ ܟ‬A26 1 7‫ ܐ ܟ ܙ‬17‫ܟ‬ nn an IA2905600522 ‫ܘܟ‬ 2‫ ܙ ܘ‬8‫܀ ܙ‬ ‫ܙ‬7‫ܟܐ‬ 171‫ܟ‬

SEQUENCING ROUTINE page 4

00 60 1828 0560 0000 00 34 1815 1715 noon 00 401A1A0000 1716 50 26 on 1790 Onon
on 34 1816 0000 1717 96 40 1817 0000 1718 96 40 0000 1737 1719 ‫ܘ‬ ‫ ܘ ܟܟ‬nno 17 ?? 1‫ܐ‬2‫ܘ‬
0 ‫ ܘܟ‬18 19 ‫ܘܪܘܪ ܐܙܐܐ‬ 56 ‫ ܝܐ‬6‫ܟ‬66 1 ? ‫ܘܘ‬ 1721 52 22 1716 0000 0000 5m Onno 1790 1722
‫ܟܟ‬ ‫ ܟܟ‬71‫ܕ‬8 281‫ܟܕܟܘ ܟ‬ ‫ܟܟ‬ ‫ ܟ‬1719 ‫ܘܪܟ‬ ‫ܟܟܟܟ‬ ‫ܗܘ‬ ‫ ܟܝ‬71‫ܕ‬8 1 ‫ ܘ‬24 ‫ܙ ܟ ܟ ܟ‬ ‫ܟܘ‬ ‫ ܕ ܟ‬7‫ܘܙ‬ ‫ ܕ‬28‫ܟܟܟܟ ܙ‬
‫ܟܟ‬ ‫ ܟ‬n 1‫ܟ‬2 ‫ܐ‬28‫ ܟܘ ܙ‬56 nn 40 1718 1822 0557 nn 20 nnnnnnnnnnnn ‫ܘܟ‬ 24 ‫?? ” ܕ‬ ‫ܪܙܘܘ ܟܘܟܟ‬
02 on noon noon 0000 ‫ܪܪ‬ ‫ܕܕܕܕ ܕܕ‬ ‫ ܝܕ ܙ‬8 into ‫ܟܟ‬ ‫ ܬܕ‬71‫ܟܙ ܕܕ‬0‫ܘܟܟ ܟ‬ 50 24 nnnn JA201722
60 An tanq qnon in nonnc nnnn non nonnnnnnnnnnnnn ‫ܘܟ‬ 2 ‫ܕܐܙ ܘ‬1 ‫ ܟ‬non nonn
59 20 02AA0000 1724 on 10 000 000 000 onnononnnnnnn ‫ܟܪ‬ 6 ‫ ܙ ܘ‬R 3‫ܙ‬ 156 ‫ ܐ‬ons
00 40 0559 1732 0559 00 72 AAA 1733 7567 no 60 174 A 1714 0569 no no onn3 anno 0003
0 60 0568 nnnn 1972 ‫ܐܙ‬ ‫ ܬܐ‬6569 ‫ܘ ܘܘܘ‬888 50 22 1713 0887 anno 50 26 000 1714 1715
‫ ܘܟ‬28 18 12 1‫ܐܕ‬5 ‫ܟܟܟܟ‬ on 67 0575 non nonn ‫ܘܪ‬ 42 1113 1 A ‫ܟܐ‬ ‫ܟܘܘ‬ ‫ܟܟ‬ ‫ܟܝ‬ 1 ‫ ܐ‬13 ‫ ܙ ܟܟܕܟ‬715
no 42 1715 1825 onon no 72 1825 0000 n5 A no 40 1829 1732 0560 t, ‫ܟ‬ s 26 1‫ܐ‬51 171‫ܟ‬
00 60 0000 0000 0582 on 40 1828 1732 0560 ‫ܘܟ‬ 26 18 1 ‫܀‬ ‫ ܐ‬71 1715 ‫ ܘܘ‬3 ‫ ܐ ܟ‬715 1815 1718
0 40 1818 0000 1716 0 60 000 Onnon534 ‫ܟܘ‬ ‫ ܘܟܟܘ‬1732 6559 50 24 1724 1725 1715
50 28 1748 1715 0000 on 63 0590 0000 0000 00 60 1785 1715 0000 00 42 1715 TR25 0000
00 70 1825 0000 0594 ‫ܘܟ‬ 20 ‫ ܟܠܐܐܐ‬18 14 1715 no 60 non Onn 595 5‫ ܪ ? ܟ‬1 ‫ܙܕ‬ 1826 171‫ܟ‬
nn 34 TA15 nnnn 1715 ‫ܘܘ‬ ‫ ܟ ܕ ܢ‬HA5 598 ‫ܕ‬8 85 no Annnnnnnnn0509 ‫ ܙ ܕ‬11 6111 1111 111 ‫ܐ‬
‫ܐ܀ ܟܘ‬ ‫ | ܙ‬1‫ܕ‬ 1735 ‫ ܘ‬non nn 40 NAA5 on 0885 no 42 1775 1825 Onn ‫ ܟ | ܟܘ‬1 28‫ ܟ‬6‫ܟ‬66_5661
50 20 TA14 1725 1725 on 600000 Oon n6n6 50 24 1725 1826 1725 502A 1749 1725 0000
00 63 0609 non non on 6‫ ܘ‬1785 ‫ܐ‬7‫ܕ‬5 ‫ܟܟܟܟ‬ 34 1 8 1 ‫ ܟ‬1725 ‫ ܕ‬725 ‫ ܝܟ‬6‫ ܟ‬18 86 non cohn
‫ܘ‬8 60 nnnn 1886_nnon ‫ܘ‬ ‫ܙܟܕܕ ܐ ܕܟ‬ 60 61 6886 000 0886 ‫ ܪ ܟܟ‬n naan non n ‫ܟܘ ܀‬
08 0650 onionon ‫ܟܘ‬ A 6 168 ‫ܟܪܟܟ ܕܪܪ ܐ‬ 00 80 16 88 003 0000 ‫ ܙ ܘ ܘ‬R R ‫ܘܘܘܟ ܬܟܟ‬
00 80 1684 0005 0000 00 80 1685 0006 non no_83 1733 ‫ ܘ‬o7 0011 00 70 1704 733 1630
on 6‫ܘ‬ 1832 ‫ ܦܪܘ‬1 ‫ ܘ‬8 ‫ܟ ܛ‬ ‫ܘ‬ 91 ooooooqqt non : 61 98000n ononnon ‫ ܟܘܟ ܐܘ ܕܪ‬n onto_anna
04 91 0000 0000 0000 099A Ono Onnnnnn ݂‫ ܟܟ‬8‫ܐ ܘ‬16 A noni nnn 1 00 60AA0615 0000
no 60 1839 0621 0615 on 42 OAA6 1727 nonn 00 72 1791 0000387 11 61 ‫ ܟ‬RA6_9566 Anna
99 6n onnn OA86 0787 on 59 anon nnnnnnn ‫ ܬܪ‬42 ‫ ܘܪܪ‬1 ; nong_31 ‫ ܟܟ‬2‫ ܟܘܩ ܀‬2 17 3 ‫ܟܘܟܘ ܟ‬
‫ܕ ܀ ܟܟ‬ ‫ܘܟܘܘ‬ 0‫ܪܪ‬8 ‫ܟܟܟܟ‬ on 62 nnnnnnnnnn 0 61 nnnnnnnnnnnn no ‫ܟܟܟܟܟ ܘܟ‬non nnon
04 42 onon 1731 00A on 42 0062 1730 non on 41 noon 000Anno 00 63 006 0641 0641
‫ܕܟ ܘܪ‬ ‫ܘܟܪܘ‬ ‫ܟܘܟܘ ܟܟܟܟ‬ 04 42 0000 1:31 0308 no no non non non ‫ܘܘ‬ ‫ܘܘܘܘ‬ ‫ܘܘܘ‬ ‫ܘܘܘܘ‬

1076‫ܟܟ‬input-
‫ܟܘ‬
output locations inserted here .
‫ܟܟܟܟ ܕܟܟܟ ܪܟܕܟ‬ 39 ‫ ܟܟ ܕܪ‬na non conn ‫ܪܙܙܙ ܟܟܟܟ ܟܟ ܟܘ‬
‫ܟܟܟܘ ܟܟܟܟ ܙܙܙܙ ܘܟ ܕܘ‬ ‫ܟܟ‬ ‫ܟܟܟܟ ܟܟ‬ ‫ܟܟܟܟ‬ ‫ܙܟܟܘ‬ ‫ܟܟܪܟ ܟܕ ܟܟ‬ ‫ܟܟܟܘ‬ ‫ ܟ‬non ‫ܘܘ‬ ‫ܘܘ‬ ‫ܘܘܘܘ‬ ‫ܘܘܘܘ‬ ‫ܘܘܙܙ‬
‫ ܟ ܟܟ ܟܟ ܟܟ ܟܟ‬non ho11 on nnnnn0 0000 non 7 ‫ܟ‬ ‫ܟܟܟܟ ܙܟܟܘ ܟܟܪܟ ܟܕ‬ ‫ ܬܘܟܪ ܘܪ ܘܘ‬ad as non
04 580001 0000 0060 on 60 4684 0892 0000 h ‫ ܟ ܛ‬4 noon dood onT1 1‫ ܪ‬6‫ܝ ܟ‬86‫ܘ ܝ‬888 non
‫ ܟ ܟ ܘ‬R -oni_ton
‫ܘ‬ 0‫ܘܟ‬ on 600000 nano 2000 11 ‫ܘܩܪܘ ܟܘܟܟ ܟܘܘܟ ܘܘ‬ ‫ ܘܘ ܘܘ ܘܘ‬n nonn 11‫ܙ‬
an on nno a nnno qnon n‫ܟ‬ ‫ ܟ‬1 ‫ ܟ‬non nonn nnr an on anno 1nna non ‫ ܘܘ‬8‫ ܟ‬1989 071‫ܪܘܘܘ‬
‫ ܟܘ‬6 ‫ ܐ ܟ‬7‫ܘ‬1 ‫ܘ‬989 nonn nn 6 ‫ܪܘܕܟ‬ ‫ ܙ‬7‫ܪܘ‬ ‫ܟܟܟܟ‬ ‫ ܘܟ‬An non ngga non ‫ ܘ ܙܟܟܟ ܟܟ ܟܟ‬non ‫ܪܟܕܟ‬
on 60 nnnn nonn ‫ܘܟܐ ܘ‬ 68 ‫ ܟ‬8 ‫ܙܘܟܘ‬ ‫ܘܙܘܟ‬ ‫ܕܕܙܟ‬ on 60 1767 016 0797 OR 5Annon on 0120
an on toon and dona 0? 91 0652 0652 0001 no 60 noon onnnn06 ‫ ܟܪ‬6 ‫ܘ‬ ‫ ܀ ܐܘ ܝ ܟܐ‬4 ‫ܟܟܟ‬
on an onon nonnnnnn ‫ܟܟ‬ ‫ܟܟ‬ ‫ܐܠܐܟܟ‬ ‫ܐܐܐܐ‬ ‫ܟ ܐܐܐ‬ ‫ ܙܘ‬: ‫ܘܘܘܘ ܘ‬ ‫ܕ ܕ ܟܘ ܘܟܟܘ‬ on 42 AR6 1764 IN A 4
‫ܟܟ‬ ‫ܟܟ‬ ‫ ܟܟ‬8‫ܕ‬ 1‫ ܩܘ‬5 ‫ܟܟܟܟ‬ ‫ܟܘ‬ ‫ܟܟ‬ ‫ܙܟܟܟ‬ ‫ܙܟ ܟܟ‬ ‫ܟܟܟܟ‬ ‫ | ܟܟ ܟܟ‬7‫ܟܟ‬ ‫ ܙܟ‬7‫ܟܟܟܟ ܢ‬ ‫ܘܪ‬ ‫ܟ ܛ‬ ‫ ܟܟܪ‬n nnn ‫ܟ‬ ‫ܟܕ ܙܟ‬
96 60 1691 0000 0000 08 58 0003 0000 0179 50 20 0892 000 1701 on 42 1084 1764 1706
‫ܘܘ‬ 6 ‫ ܟܪ‬n anno 21‫ܝ‬1 A 5A0001 Onnon306 nn_on_nah1 0001 ‫ܟܟܟܟ‬ on 42 ORA 1764 TR4
‫ܢܟ‬ ‫ܟܟ‬ ‫ܛܟܟܘ‬ ‫ܟܘܟܘ ܟܟܟ‬ na honon 1993 nnnn ‫ܟ‬ T ‫ܟ ܟ‬8 8‫ܝ‬ 671‫ ܟ ܟܐ‬3‫ܘ ܙ‬ nn 6‫ ܘܪܪܬ ܘ‬ad as anno

EXHIBIT XI

70
 SEQUKNCING ROUTINR page 5
00 60 0000 0892 0000 04 58 0001 0000 0342 99 09 9000 0000 0000 00 60 0000 0988 0000
04 58 0001 0000 0352 00 60 0000 0000 0368 00 60 0000 0000 0372 9A 05 0000 0000 0000
00 00 0000 0000 0009 A9 01 0000 0000 0000 00 60 0988 0884 0000 00 60 0000 0000 0248
00 60 0000 0000 0260 00 60 0000 0000 0252 00 60 0000 0000 0112 08 58 0001 0000 0408
CC 60 0892 0000 0000 08 58 0001 onoo n420 00 60 0891 0000 0000 01 91 OA84 08840001
06 58 0003 0000 0457 50 22 0000 0989 1701 50 20 0000 0990 1702 00 60 1744 0476 0479
00 60 0000 0000 0297 00 60 1795 0496 0000 00 60 1796 0241 0453 00 60 1811 0496 0398
00 60 0000 0000 0373 00 04 0000 0000 0000 00 03 9900 0000 0000 97 05 0000 0000 0000
90 01 0000 0000 0000 AR 01 0000 0000 0000 00 00 onon onon 1849 10 00 0000 0000 noon
12 00 0000 0000 0000 50 22 0000 0000 0000 50 24 0000 1 822 0000 00 02 0000 0000 noon
50 20 0000 0000 0000 50 20 0000 0000 0388 00 10 0000 0000 0000 97 15 0000 0000 0000
99 15 0000 0000 0000 50 20 1830 0000 1724 50 20 1827 0000 1724 99 05 0000 0000 0000
00 60 0000 0000 0585 00 60 0000 0000 0620 98 15 0000 0000 0000 00 42 0000 1764 0000
00 40 0000 1726 0000 00 40 1084 0000 1084 11 42 0000 1840 1728 00 80 1686 0001 0000
00 83 1727 000 7 0624 01 111000 0000 0000 02 42 0000 1818 0000 00 70 1733 0000 021A
50 22 0000 1728 1701 50 26 1701 0000 1702 50 24 1702 0001 1702 00 60 1701 0000 0000
00 60 1702 0001 0000 00 60 0000 0000 0166 0000 0000 0000 0000 99 00 1994 0000 0000
99 00 3983 0000 0000 99 00 5962 0000 0000 99 00 7926 0000 0000 99 00 9871 0000 0000
99 01 1791 0000 0000 99 01 3683 onononno 99 01 5542 0000 0000 99 01 7364 0000 0000
99 0 2 4215 0000 0000
99 01 9146 0000 0000 99 02 0884 0000 0000 99 02 2575 0000 0000
99 02 5804 0000 0000 99 03 0114 0000 0000 99 02 8814 0000 0000 99 03 0234 0000 0000
99 03 1594 0000 0000 99 03 28 94 0000 0000 99 034134 0000 0000 99 03 5314 0000 0000
99 03 6432 0000 0000 99 03 7493 000 0000 99 03 8492 0000 0000 99 03 9435 0000 0000
99 04 0320 0000 0000 99 04 1149 0000 0000 99 04 1924 0000 0000 99 04 2647 0000 0000
99 04 7319 0000 0000 99 04 394 3 0000 0000 99 04 4520 0000 0000 99 04 5053 0000 0000
99 04 6784 0000 0000
99 04 59 43 0000 0000 99 04 5994 0000 0000 99 04 6407 0000 0000
99 04 7942 0000 0000
99 04 71 28 0000 0000 99 04 7441 0000 0000 99 04 7725 0000 0000
99 04 8214 0000 0000 99 04 8610 0000 0000 99 04 8778 0000 0000
99 04 8422 0000 0000
99 04 9286 0000 0000
99 04 8928 0000 0000 99 04 9061 0000 nono 99 04 91A 0000 0000 04 95 98 00000000
99
99 04 93 79 0000 0000 99 04 9461 0ono 0000 99 049534 0000 0000 99 04 9741 0000 0000
99 04 9653 0000 0000 99 049702 0000 0000 99 04 9744 0000 0000
99 04 9886 0000 0000
99 04 9813 0000 0000 99 04 9841 0000 0000 99 04 9865 0000 0000 9942 0000 0000
99 04
99 04 9903 0000 0000 99 04 9918 0000 0000 99 04 9391 0000 0000 99 04 9972 0000 0000
99 04 9952 0000 0000 99 04 9960 0000 0000 99 04 9966 0000 0000
99 04 9987 0000 0000
99 04 9977 0000 0000 99 04 9981 0000 0000 99 04 9984 0000 0000 99 04 9994 0000 0000
99 04 9989 0000 0000 99 04 9991 0000 0000 99 04 9993 0000 0000 99 05 0000 0000 0000
99 04 9995 0000 0000 99 04 9996 0000 0000 99 05 0000 0000 0000
00 00 0000 0000 0000
00 40 1946 0000 1938 92 60 0000 0000 0000 00 40 1947 0000 1934 0000 0000 0000 0000
00 00 0000 0000 0000 00 00 000 0000 0000 00 00 0000 0000 0000 00 40 1945 0000 1934
00 61 0000 0000 1940 12 60 1936 1935 0000 88 42 1946 0000 0000 00 59 9999 9132 0001
00 60 1944 1937 1934 00 68 1939 1937 1935 A8 40 1935 1946 1935 01 40 1968 0000 1952
00 59 9998 1936 1940 00 40 1967 0000 1960 91 60 0000 0000 0000 12 42 1945 1949 0000
00 60 1971 1962 0000 00 00 0000 0000 0000 on 6A 0000 0000 0000 0000 0000 0000 0000
88 40 1970 0000 1956 00 00 0000 0000 0000 00 41 1956 1969 195A 00 00 0000 0000 0000
00 68 0000 0000 1962 00 00 0000 0000 0000 00 61 1965 0000 1971

SEQUENCING ROUTINE pago 6

00 40 1965 1970 1964 00 00 0000 0000 0000 00 00 0000 0000 0000 00 60 1961 1962 195 ?
00 59 9999 9140 0001 05 99 9981 9650 0000 00 01 0000 0000 0000 00 02 0000 0030 0097
00 60 1956 1962 1962 00 60 0000 1997 0000 08 60 0000 0000 0000 00 40 1993 0000 1991
92 60 1997 0000 0000 00 40 1994 0000 1978 16 60 0000 1997 0000 00 00 0000 0000 0000
00 63 0000 1991 1992 00 40 1996 0000 0000 03 60 0000 0000 0000 11 34 1995 0000 0000
86 60 0000 0000 0000 00 40 1996 0000 0000 00 40 1997 0000 1997 50 36 1978 0000 0000
50 32 0000 1997 0000 50 34 1995 0000 0000 50 20 1997 0000 0000 00 72 0000 1997 1995
00 00 0000 anno 0000 60 61 0000 0000 1991 no on 1997 0000 0000 50 20 0000 0000 1978
99 05 0000 0000 0000 67 01 0000 0000 0000 00 00 0000 0000 0000 02 91 0051 2040 0001

EXBIBIT XI

71
sequence , it first calculates the value for TL - Tr , puts will be enhanced if they are presented in an
the slack . This is stored in the event file but is order which allows easy reference to selected
not needed on the working tape . After calculation , events . The input to Run 09 is the same computed
the value of TL - Te is tested by subtracting 5.0 output event tape in ascending order of the event
from it . If the result is zero or negative , a 9 code code . The program itself is almost identical to
is generated and stored in the record code position . Runs 03 and 04. Further discussion of sorting pro
The existence of a value for Ts is then tested if grams is found there .
present , the values ( 2 ) Output Edit (Run 10 ) . The output edit run is
Ts - Te TL - Ts a necessary function for two reasons . First , the
and event file contains a great deal more information
ОТ Е OTL
than is useful on a report. Therefore, some of the
are calculated . Ts represents a measure of the information in the file should be screened out before
elapsed time from time " now " to the scheduled date . the report is printed . Seconaly , as noted earlier , in
As time “ now ” is advanced, the Ts or elapsed time order to maintain a reasonable amount of precision
must be reduced . This is adjusted by manually in the computations, it is necessary to use a full
storing the time differential from the last reporting NORC word and floating point arithmetic operations .
date , or the difference from time zero for the first In reporting values , however , it is only reasonable
run after setting up the system . This adjustment is and consistent to report Te and Tl to one decimal
applied to every Ts in the file . The adjusted value place . The inherent accuracy of the data is no
is stored in the event file after each computation better than this level , and it is more readily under
of TL stood in this format . The output editing routine
These values are looked up in a previously accomplishes both these results .
stored table of the normal probability function . The The input to the output edit run is a computed
resulting probability expression for event file tape . The resulting output is an edited
Ts - TE output tape containing all events and a second
OTE edited out put tape containing all events having a
report code . A report code can be inserted into the
is stored in the event file . The product of the two file through the updating routine or it can be gener
values resulting from the table look -up is also ated internally as a result of the critical slack
stored in the event file .
calculation . The format of both tapes is identical ,
The IlL calculations follow in a similar manner using a four word event message . The output tapes
for each event in the event file . The reverse calcu are then ready to be converted to cards on the CTC
lation is based on the successor events rather than operation . It should be noted that these cards can
the predecessor events , since these are the prede be sorted and manipulated using the IBM 083 card
cessor events in the reverse sense . The slack
sorter . Thus , the printed reports can be prepared
( TL - Te ) and the probability calculations are in any order .
carried out as indicated above . Adjustments in the The program operations are straightforward . A
final Tg and oi e values are made for the activities
E E block is read into memory from the input tape . The
leading up to time now as in the earlier Te calcu program extracts the first event , and sets up the
lotions . event code in word 1 of the output tape storage .
A detailed program flow chart and the absolute The event code is also placed in word 1. The pro
coding for Run 08 are found in Exhibits XII and gram then selects the value for Te , converts it to
XIII , following this page . The generation of the fixed point form , rounds off the number to one
output tape from Run 08 represents the end of the decimal place , develops the coded sign of the
computer calculations. It is now necessary to trans value , and places the entire result in the last five
late the findings to a usable form . digits of word 1. In a similar manner , it constructs
the correct form for values of TL , TL - Te , and
2. OUTPUT PROVISIONS
Ts placing them in word 2. The probability value
( 1 ) Output Surt ( Run 09 ) . Run 09 , the output and the schedule date are placed in word 3. Zeros
sort run , has two functions . First , it re -sorts the (NORC blanks ) are placed in word 4 , and the entire
event file tape into exact event order and prepares result is placed on the output tape . The program
an input tape for use in the file updating run ( Run selects the next event , and the cycle is repeated
06 ) . Generally , when the system is set up and until all events have been edited and placed on the
operating for large networks , the utility of the out output tape .

72
 PROGRAM FLOW CHART
COMPUTATION RUN #08
S. 180

READ IN TAPE LABEL COMPARE FRED CODE

* TH EVEN FROM
STORAGE

LOOP 0

KEYINK CONSTANTS 5.185

RUNNINDATE AND STOP COMP . LONGEST PATH
TE TE

SELECT LARGEST TE
13
COMPARE LABEL WITH TE
KEYED IN DATE 5. 190

WRITE SON 04 READ

BACK 2 BLOCKS READ
FOR MARDI BLOCK SET
EXITS S080 AND 5. 110
PRINT
STOP

EXIT
CLEAR MODS . EOF
ON 02 TAPE
5.090
5.080

5
5.150
READ 01 INPUT TAPE
TCF

3.192

FAD I BLOCK ON 04
YEPE SET EXTS 5.090 ,
EOF 700
5 , COUNTER
NO

EOD STORAGE
COMP . BLOCK LENGTH 5. 10
CLEAR M

5.

SENSE , 03 COUNTER
-BLOCK LENGTH : 0

5. 110

003 COUNTER
RESETEE STO

5.100 RESET EXIT IN BOX

5 152 AND 5 105

STORE EVENT INDEX :

FXTRACT AND SENSI
PRED , INDEX = 0

= 0

COMP ' ' AND STO

K FOR AD JUSTING .
AND STORE IN
FILE AS .."
SET EXIT 5.700

$ 150

STORE AND WRITE

( 200 WORDS FOR
FORWARD COMP .
WRITE O2 TAPE

5.250
STORE IN S. EVENT CODE
TE AND TE CRITE
WHEN STOPE 200
EVENTS ON 04 TAPE

S. 103 5. 110

5.01 EXIT S. 100

5.070

EXHIBIT XII
5.105
1
 COMPUTATION ROUTINE Page 1
-02 91 0051 0056 0000 09 96 0008 1753 0001 00 6A 0000 0000 0006 00 60 0000 0000 0008
60 ბი ბიბი ბიიი ძილი 00 61 0000 0000 0001 0000 0000 0000 3952 02 91 0051 0056 000n

-02 91 0051 1796 0001 01 94 1602 1602 0000 08 68 0009 0000 1713 ბ ) 61 გიგი 88
00 40 1691 1697 1611 00 70 1602 0000 0014 00 61 0000 0000 0000 00 40 1710 1697 0000
00 40 1629 0000 1629 03 91 1629 1629 0000 08 60 0017 0000 1730 00 57 0000 0000 0000
01 94 0397 0000 0001 OA 68 0020 0000 1713 00 67 0000 0000 0086 00 60 Onno 1603 onon
04 42 0000 1637 1604 08 42 1603 1638 0000 00 41 1604 0000 1605 00 40 1639 0000 1605
00 52 0000 0000 0000 00 41 1605 1606 0000 00 63 0035 0031 0000 00 61 0000 0000 0000
00 60 1658 0050 0000 00 40 0019 1639 0019 00 60 1640 1606 0000 00 60 1645 0036 0019
13 60 6397 1607 0000 00 60 0403 1608 0000 02 42 160A 1642 1609 00 42 160A 1643 1610
00 72 1609 1640 0044 00 60 1646 0054 0000 00 70 1610 1640 0044 00 60 0000 0000 0045
00 60 0000 0000 0198 on 60 1647 0054 0000 96 60 1607 0000 0000 00 40 0050 0000 0050
00 54 0000 0000 0000 00 60 6397 8598 0000 00 50 0000 0000 0001 06 58 0001 0000 004 B
00 40 1607 1606 1606 92 60 1607 0000 nono 00 40 0036 0000 0036 ბ) 60 ბიიი ტიიტ ბი 24
00 40 1611 1607 0000 00 41 1650 0000 0000 00 63 0000 0060 0059 OC 50 1647 0069 0062
00 60 0058 0079 0070 00 60 1647 0079 0000 00 60 1646 0069 0062 00 51 გიგი ჩიიი ბიბი
96 60 1607 0000 0000 00 40 1649 0000 0068 00 60 459A 0794 0000 00 40 0065 164 A 0065
00 40 1611 1639 1611 00 00 0000 0000 0000 no 00 0000 0000 0000 96 60 1611 0000 0000
00 41 0000 1648 0000 00 40 1652 0000 0073 00 00 0000 0000 0000 OA 60 0074 0000 1730
00 60 0000 0000 0000 00 60 1651 0065 0000 00 40 1652 1639 1652 00 60 1640 1611 0000
00 00 0000 0000 0000 01 98 0000 0000 0000 02 91 0000 0000 0000 02 9A 0000 nonn 0000
00 57 0000 0000 0000 00 40 1681 1652 1652 00 60 1640 1606 0095 00 70 1611 1640 0080
00 60 1646 0079 0070 00 80 1602 0001 0000 00 80 1631 0002 0000 00 80 1629 0003 0000
00 80 160 0004 0000 00 80 0000 0005 0000 00 80 0000 0006 0000 00 83 0000 000 7 0011
02 94 0397 0000 0000 08 68 0096 0000 1713 no 67 0000 0000 0311 00 60 0000 1603 0000
04 42 0000 1637 1604 OR 42 1603 1638 0000 00 41 1604 0000 1605 00 40 1639 Onno 1605
00 51 0000 0000 0000 on 60 0000 0000 0000 01 94 0999 0000 0001 ი 8 68 616 იიიი 1713
00 67 1653 1634 0348 00 60 0000 1613 0000 00 40 0105 1639 0105 00 60 1653 1621 0000
04 42 1613 1637 1604 08 42 1613 1638 0000 00 41 1604 0000 1614 00 40 1639 00no 1614
00 52 0000 0000 0000 00 60 1646 0140 0000 00 60 0000 0000 0000 00 41 1605 1606 0000
00 63 012 0121 000 on 61 0000 0000 0000 00 60 1640 1606 0000 00 60 1645 0124 0000
00 60 1656 0104 0095 00 60 0403 1608 0000 13 60 4397 1607 0000 02 42 1608 1642 1609
00 42 1608 1643 1610 00 72 1640 1609 0131 00 72 1610 1640 0131 00 60 1699 0140 0133
00 72 1610 1640 0133 00 60 1693 0140 0133 96 34 1609 1644 1609 96 34 1610 1644 1610
00 42 1609 1711 1609 ხი 42 161 0 1711 1610 92 60 1607 0000 0000 00 40 0124 0000 0124
00 40 1607 1606 1606 00 60 0000 0000 0000 00 41 1614 1615 0000 00 63 0149 0144 0000
00 61 0000 0000 0000 on 60 1640 1615 0000 00 60 1659 0149 0000 00 60 1657 0117 0000
00 60 1660 0167 0000 00 60 0000 0000 0105 00 60 1005 1616 0000 13 60 6999 1610 0000
02 42 1616 1642 1617 00 42 1616 1643 1618 00 72 1617 1640 0156 00 60 0155 0169 0156
00 60 1646 0169 0141 96 34 1617 1644 1617 96 34 1618 1644 1618 92 60 1619 onno onon
00 40 0149 0000 0149 ტი 42 1617 1711 1617 no 42 1610 1711 1618 00 40 1619 1615 1615
96 60 1619 0000 0000 00 40 0167 0000 0167 00 60 6999 0598 0000 00 40 0165 1548 0165
06 98 0001 0000 0165 00 60 1661 0165 0000 00 60 0000 0000 0000 92 60 1619 0000 0000

COMPUTATION ROUTINE page 2

00 40 1672 0000 0000 00 40 1694 0000 0174 00 42 0599 1655 1620 00 00 0000 0000 0000
00 41 0000 1620 0000 00 63 0000 0178 0207 00 60 0197 0196 0179 00 60 1640 1635 0227
00 40 1611 1619 0000 00 41 1650 0000 0000 00 63 0185 0182 0183 00 60 0196 0206 0184
00 60 1700 0206 0198 00 60 0043 0196 0000 00 54 0000 0000 0000 96 60 1619 ბიიი იიიი
00 40 1662 0000 0194 on 42 0599 1655 0000 00 72 1636 0000 0191 00 61 Onno Onno 0000
00 60 8598 1200 0000 00 40 0191 1648 0191 00 40 1639 1611 1611 00 00 0000 0000 0000
00 42 0399 1655 1636 06 00 6000 ბიბ იბიც no 60 0000 0000 0141 96 60 1611 0000 0000
00 41 0000 1648 0000 00 40 1664 0000 201 00 00 0000 0000 0000 OA 60 0202 0000 1730
00 40 1664 1639 1664 60 60 1640 1611 000 ) 00 60 1663 0191 0000 26 იი ბიიი იიიი იიბი
00 72 1653 1621 0373 00 60 1704 0148 0000 00 60 0177 0169 0177 00 42 0201 1653 0000
00 72 1666 0000 0218 00 40 1664 1653 1664 00 60 1701 0338 0000 00 40 1683 1666 0223
00 40 1684 1666 0224 00 40 1684 1666 0343 00 60 0000 0000 0223 00 41 1664 1653 1664
00 40 1653 1701 0338 00 40 1683 1688 0223 00 40 1684 1688 0224 00 40 1684 1688 0343
00 00 0000 0000 0000 00 00 0000 0000 0000 00 60 1689 0148 0000 00 60 1640 1636 0144
00 42 0605 1655 1620 00 42 4398 1655 0000 00 40 1644 0227 0227
00 70 1620 0000 0234
00 40 1707 1624 1624 00 72 0000 1617 0227 00 60 1640 16 24 0000
00 61 0000 0000 0000
00 60 1669 0227 0000 00 40 0599 1653 0599 02 42 0604 1542 0000
14 60 0000 1632 0000
001 იიცი 1632 ბიიი 00 63 0000 0282 0251 00 60 1667 0272 0000
00 60 0177 0250 0000
02 60 1653 0000 0000 00 40 4398 0000 4398 00 42 4403 1643 0000
12 42 4398 1671 1632
00 41 0000 1632 0000 00 63 0000 0253 0252 00 00 0000 0000 0000
00 40 1672 1644 1672 42 4403 1643 0000
00 61 0000 0000 0000 00 61 0000 0000 0000 AA
92 60 1698 1672 0000
00 41 4398 0000 4398 00 40 1625 1607 0000
00 41 1650 0000 0000
00 63 0000 0260 0261
00 60 1646 0271 0262 00 60 1678 0271 0262 ბ ) 54 ბბბბ ბიიბ ბიბი
96 60 1607 0000 00 60 0259 0281 0273
00 40 1612 1639 1612 0000
00 40 1674 0000 0270
00 60 4397 1401 0000
00 40 0266 1648 0266 00 40 1625 1639 1625 00 00 0000 0000 0000
0000 0000 0000 0000 00 50 0001 0000 0000 00 41 0000 1648 0000
00 00 0000 0000 0000
00 00 0000 0000 0000 96 60 16 25 0000 0000 00 40 1677 1639 1677
00 40 1677 0000 0276 08 60 0277 0000 1730
00 60 1640 1625 0000 00 60 1673 0266 0000 00 60 0197 0250 0000
88 42.0604 1642 00 00 0000 0000 0000 00 40 1619 1626 0000
00 60 1699 0272 0000 0000
On 63 0000 00 41 0599 0000 0599 00 60 0301 0310 0000
00 41 1650 0000 0000 0290 0292
00 60 0289 0310 0302 00 60 1646 0300 0293 96 60 1619 0000 0000
00 60 1682 0300 0293 00 54 0000 0000 0000
00 60 8598 0798 0000 00 40 1626 1639 16 26
00 40 16 79 0000 0299 00 40 0296 1648 0296
0000 0000 0000 0000
იი 00 იბბი ბიიი ი 000 96 60 1626 0000 0000
00 40 1652 0000 0305 00 60 0000 0000 0243 0A 60 0306 0000 1730
00 41 0000 1648 0000 00 00 0000 0000 0000
00 60 1640 1626 0000 00 00 0000 0000 0000
00 40 1652 1639 1652 00 60 1680 0296 0000
00 60 1686 0310 0302 00 72 1681 0000 0316
00 70 16 26 1640 0313
00 42 0095 1653 0000 00 40 1630 1653 1630 00 40 0095 1681 0095
00 60 0000 0000 0395
00 91 1652 1681 1652 00 40 1685 1683 0332 00 72 1681 0000 0325 00 40 1685 1684 0334
00 60 0333 0336 0332 00 40 1681 1684 0333 00 40 1652 1681 1657
00 40 1681 1683 0335 on 40 1684 1685 0333 60 41 0095 1681 0295 00 40 16A3 1685 0335
00 40 1683 1681 0332 00 00 0000 0000 000 00 40 1684 1681 0334 0000 0000 0000 0000
00 60 0333 0336 0332 on 00 0000 0000 0000 00 00 0000 0000 0000 ბი ბი იიიი იიიიიიიი
00 00 0000 0000 0000 00 67 1688 1628 034 3 00 60 0000 0000 0121 00 60 1666 1621 0111
08 68 0339 0000 1713 00 60 1685 1634 0000 00 60 0000 00 00 0000 0000 0000
1613 0000
00 00 0000 0000 0000 00 60 1708 0105 0381 00 60 1691 0148 0144 01 98 0000 0000 0000
00 00 0000 0000 0000
03 91 0000 0000 0000

EXHIBIT XIII

75
 COMPUTATION ROUTIE page 3
02 98 0000 0000 0000 03 98 0000 0000 0000 04 98 0000 0000 0000 05 98 0000 0000 0000
06 98 0000 0000 0000 09 98 0000 0000 0000 00 60 0013 0001 0001 00 00 0000 0000 0000
00 70 1611 1640 0210 00 60 1712 0206 0198 00 72 1628 1688 0365 00 60 1640 1628 0000
00 60 1646 0169 0000 00 60 1689 0148 0359 00 60 1706 0345 0338 00 80 1629 0001 0000
00 80 1640 0002 0000 00 80 0000 0003 0000 00 80 0000 0004 0000 00 80 Onoo 0005 0000
00 80 0000 0006 0000 00 83 1612 0007 0349 00 60 1668 0148 0000 00 60 0177 0169 0177
00 59 0000 0000 0000 04 42 0000 1637 0380 00 41 1650 0000 0000 00 63 0165 0000 0000
00 61 0000 0000 0000 00 00 0000 0000 0000 00 72 1634 1640 0383 00 61 0000 0000 0000
00 40 1635 1639 1635 00 41 1707 0000 0000 00 63 0362 0362 0000 00 60 0174 0387 0000
00 00 0000 0000 0000 00 80 0000 0001 0000 00 80 1629 0007 0000 00 BO 1612 0002 0000
00 80 1620 0003 0000 00 80 1640 0004 0000 00 80 0000 0005 0000 00 81 0000 0006 0387
00 70 1625 1640 0366 00 60 1709 0281 0273 00 00 0000 0000 0000 00 00 0000 0000 0000
1236 input -output locations insorted bero .
0000 0000 0000 0000 00 00 0000 0000 0000 00 00 0000 1111 0000 00 00 1111 0000 0000
00 00 0000 0000 0001 00 00 0000 0000 0000 00 00 0000 0000 0004 00 00 0000 0000 1100
00 00 0000 0000 0011 00 00 0003 0000 0000 00 60 0403 160A 0000 00 60 0000 0000 0000
00 60 0000 0000 0028 00 00 0000 0001 0000 04 58 0001 0000 0065 0000 0000 0000 0200
00 60 4598 0798 0000 02 91 0798 0798 1000 01 00 0000 0000 0000 00 00 1000 0000 0000
00 00 0011 1111 1110 00 60 1640 1630 0118 00 60 1646 0169 0141 06 58 0001 nono 0048
00 60 1005 1616 0000 06 58 0001 0000 0165 00 60 6999 0598 0000 08 58 0001 0000 0191
00 60 8598 1200 0000 05 91 1200 1200 0500 00 60 0000 0000 0179 05 00 0000 0000 0000
00 60 1699 0196 0179 00 60 0000 0000 0361 00 42 0605 1655 1620 00 00 0001 0000 0000
00 11 0000 0000 0000 00 00 000 7 0000 0000 00 60 4397 1401 0000 08 58 0001 0000 0266
00 60 1657 0196 0185 00 60 1696 0196 0338 03 91 1401 1401 1000 00 60 0272 0281 0273
08 58 0001 0000 0296 00 60 8598 0798 0000 02 00 0000 0000 0000 00 60 0000 0000 0302
00 91 0000 0000 0000 00 98 0000 0000 0000 04 00 0000 0000 0000 00 60 0000 0000 0319
00 60 0000 0000 0170 06 00 0000 0000 0000 00 60 0000 0000 0338 00 95 0000 0000 0000
00 60 1640 1628 0105 00 40 0167 0000 0167 00 60 1699 0272 0256 00 42 4397 1655 0000
00 50 0000 0000 0000 00 60 0000 0000 0346 39 06 0000 0000 0000 00 00 0000 0200 000 7
00 60 1646 0140 0118 00 60 0000 0000 0185 05 94 0999 0000 0000 00 60 1703 0206 019R
00 60 0000 0000 0197 00 60 0000 0000 0359 00 60 1691 0148 0144 00 60 1705 0345 0361
00 00 0000 0000 0003 00 60 0000 0000 0383 no 60 0000 0000 0366 00 01 0000 0000 0000
0000 0000 0000 1111 00 60 0000 0000 0210 00 40 1728 0000 1720 92 60 0000 0000 0000
00 40 1729 0000 1716 00 00 0000 0000 0000 00 00 0000 0000 0000 0000 0000 0000 0000
00 00 0000 0000 0000 00 00 0000 0000 0000 00 61 0000 0000 1722 12 60 1716 1717 0000
88 42 1728 0000 0000 00 40 1727 0000 1716 00 60 1726 1719 1716 00 68 1721 1719 1717
88 40 1717 1728 1717 00 59 9999 9132 0001 00 59 9998 1718 1722 00 40 1749 0000 1742
91 60 0000 0000 0000 01 40 1750 0000 1734 00 60 1757 1744 0000 00 00 0000 0000 0000
00 68 0000 0000 0000 12 42 1747 1731 0000 88 40 1752 0000 1738 00 00 0000 0000 0000
00 41 1738 1751 1740 00 00 0000 0000 0000 09 68 0000 0000 1744 00 00 0000 0000 0000
00 61 1747 0000 1753 00 00 0000 0000 0000 00 40 1747 1752 1746 00 00 0000 0000 0000
00 00 0000 0000 0000 00 60 1743 1744 1735 00 59 9999 9140 0001
05 99 9901 7470 0000
00 01 0000 0000 0000 00 02 0000 0030 0097 00 60 1738 1744 1744
02 91 0051 1796 0001-

EXHIBIT XIII

( 3 ) Output Tape and Output Card . The output allow for the generation of additional factors that
tape format is shown on Exhibit XIV , following this may be needed for future analyses . Additional
page . Word 1 contains the event code in positions information is carried in the main event file or
2 through 10 . The report code is in position 11 . could be generated from this file by relatively minor
The sign of Te is in position 12 and Te occupies modifications in the computer programs.
positions 13 through 16. Te is expressed in weeks
and tenths of weeks . Word 2 contains the sign of The output printer has been programed to recog
Tl in position 1 and Tl in positions 2 through 5. nize a change in the first three digits of the event
The sign of TL Te is in position 6 , and TL - Te code . A change here designates a change in a
is in positions 7 through 10. Position 11 is zero major system and thus in the responsible SP Tech
( blank ) . Position 12 contains the signs of Ts , nical Branch. Upon recognizing this change , the
while Ts is in positions 13 through 16. Word 3 con printer will cause the paper to advance , and it will
tains the probability value in positions 3 and 4 , and start printing the next system on a new page . The
the schedule date is in positions 5 through 10. The printer has been further programed to effect a triple
remainder of word 3 and all of word 4 are filled with space on a page when there is a change in the
zeros . These zero positions could be used for second group of three digits . The latter designates
additional output information at some time in the a change in subsystem within a major system . The
future . capacity of this coding system has been examined
The output card is shown on Exhibit XV , on and is considered adequate for the complete FBM
page 78 , and follows exactly the same format of system .
the output tape.
The output form has been pictured in the main
( 4 ) Output Form . The output form contains only body of this report . The information processing
the information needed for currently planned evalua procedure places constraints on the magnitudes of
tion . The computer system has been designed to the numbers that can be printed on the output form .

76
 EDITED OUTPUT TAPE LAYOUT

WORD
# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

EVENT CODE RPT. +

CODE ТЕ
1
o 1 x x x x X X X X х x x 1 x х Х х х
+
+ TS ✓
TL TL - TE
2
N

х х x x .
Х x x X X .
х o x x x x х х

-
P, SCHEDULE DATE ✓
3
х х x x x x x х o o ox

o
o
o

o
Io
Io

. -

4
o

o o o o o o o 0 o o o O 0 O х

EXHIBIT XIV

These constraints are summarized below . can be assured of the accuracy and fidelity of the
Number operations at NPG .
Designation of Digits Range
( 1 ) Accuracy Control Techniques . Accuracy
Col. (2) - Te 1.0 - 999.9+
Col. ( 3 ) – TL 999.9+ – 999.9
control for the NORC -PERT system is developed
Col. ( 4 ) – Slack 999.9+ 999.9 and maintained in three ways .
Col. ( 5) – Ts 999.9+ 999.9
Col. ( 6) – P. 2 .00 .99 - Internal Control . In addition to the machine
Col. ( 7 ) Schedule Date 6 Month- Day-Year
Col. (8) 24 Available for any checks of the NORC itself , several additional
additional data checks have been added to the programs . These
checks consist mainly of tests of reasonableness
The foregoing sections have described the logic and consistency as applied to the actual file identi
of the computer program . The computer program has fication and computative data. For example , before
been carefully designed to automatically discover updating any record , the event codes of the file and
inconsistencies and errors . The following section
the transaction are verified . In a similar way , be
focuses upon the detailed checks that are the mark fore events are qualified for repositioning in the
of a mature system. sequencing operation , the event and its predecessor
are verified with the corresponding event and its
3. ACCURACY CONTROL . The general need for successor .
accuracy control techniques and procedures in a
high speed data-processing system cannot be over A reasonableness check is applied to the input
emphasized . The very nature of any such system is data before updating to determine that a < m < b and
to remove from visual and manual observation data further that
which in the normal course of handling would be
sight verified many times by many people . It is not of g < 99.9.
E
enough to establish exacting rules for processing
the data in the computer program . In addition , it is Other similar checks can be applied . Moreover ,
further necessary to establish and maintain stand the entire system should be examined periodically
ards of accuracy and control for the data flowing to consider the efficiency of current checks and the
into and out of the machine operations. possible addition of new internal controls .
For the SP PERT system , the problem is aggra — Transaction and File Control Totals . The total
vated by the fact that NPG is acting as a service transactions in each batch are counted before they
function for Sp 12. As such , PERT will occupy are transmitted to NPG . During the file mainte
less than 5% of their total time . In this position , nance run , the total transactions processed are
SP cannot be assured of a continuing close follow counted . The figure is balanced with the total
up by NPG personnel over the life of the project . transactions key-punched . Finally, this count is
It is necessary , therefore , that sufficient controls transmitted to SP with the output sheets , and should
be developed and administered by SP so that they be balanced with the total transmitted to NPG .

77
 LAYOUT
CARD
OUTPUT

ZEROS CODE
EVENT -
TE T-T LE Ts SCHEDULE ZEROS ZEROS

‫ס‬
It

+
DATE

OmN
ODMN
ODMN

4‫ססס‬
UOow
1 12 14
13 24
23
22 29 34
33 38 39 40 45
44 4647 49
48 55
54 60 61 80

28
III IV

78
XV
EXHIBIT
 A manual running control of the total events in given to the dispatch clerk . The dispatch clerk will
the file is maintained at Sp 12. A beginning total then count , verify , and initial each batch. Batches
will be combined at SP or NPG only with permis
is established and adjusted by the additions and sion of the Sn 12 liaison officer.
deletions of events . Both sequencing and comput
ing runs produce a total of events in the file for -Output data will be returned to the dispatch
clerk , where it will be logged in and the actual
control purposes. These should be balanced to each
control data entered in the dispatch book . The form
other so as to test the consistency of operation at for the NORC Data Dispatch and Control Sheet is
NPG . The computing run at NPG also generates shown on Exhibit XVII, on page 81. When it
the total number of predecessors and successors . has been verified that the batch has been cor
These totals should be equal and individually give rectly processed , the dispatch clerk will issue
the total activities current in the file . instructions to NPG to destroy the input cards .
-Tape Label Control . Each tape used in the -A running control of the number of events in the
management data- processing activity at NPG has a file will be maintained by the dispatch clerk . The
tape label which uniquely identifies it . The tape beginning number of events will be entered on the
label consists of four items in a NORC word . control sheet and additions , deletions , and com
Problem Run Sequence Week pleted event totals will be entered by date and
Number Number Date Number Number batch number . The output control data will be ex
amined and the event total determined by the NORC
XX XX XX - XX- XX XX XXX . X
will be entered in the running control . Output data
The first two digits represent the problem number will not be released until running controls and
batch controls are balanced .
the PERT problem number is 39- and will remain
unchanged during its life . In the PERT system -NPG will use the last good file for each updating
there are ten runs , as described earlier . The date
and computing run . A log or register of the tape
is a six -digit calendar date . The sequence number labels will be maintained at SP and NPG . The
is used to identify the multiple runs during a day . layout of the SP tape register is shown on Exhibit
The week number is the week count which is iden XVIII, on page 82. The corrected tape label for each
tical to the date and used in the updating and com run must be verified by reference to the output con
puting runs to establish week " now . ” Normally , the trol data before any output data is released by SP.
regular runs will be numbered 0000 , and special If special runs have been made in the period be
runs assigned numbers in sequence . Each run will tween two regular runs , the liaison officer must
produce control data showing the beginning and issue specific instructions to NPG indicating the
ending tape label for the run . A simple log of the correct file tapes to be used .
tapes and their labels for each run will protect the
validity of the output data and isolate special runs
made for experimental purposes . It is proposed that
NPG hold transaction tapes for three consecutive
updating runs before destroying the data. Com
pletely analyzed files ( after computing) will be The last two chapters have been concerned with
consolidated and held for six months . A flow chart the " hardware " aspects of the computer installation
of the NORC control system is shown in Exhibit and the programing steps taken to ensure efficient
XVI , following this page . machine runs . However , a going data-processing
system must provide for more than a well -conceived
( 2) Accuracy Control Procedures
program . In addition , SP must under write the neces
-Input data will be transmitted to NPG in batches . sary costs of the system and provide the personnel
Batches will be registered by dispatch clerk at Sp for running the system . The following chapter will
12 and the batch number assigned . The number of deal with these “ nonhardware ” elements that com
lines (transactions ) will be counted before they are prise the computer environment .
V. COMPUTER REQUISITES

1. COSTS OF OPERATION steps . This is a cost of $ 5.74 per step . Approxi

( 1 ) Program Formulation Costs . Exhibit XIX , mately 221 man -days were used in completing the
on page 83 , shows a break down of the costs programing effort. This is a rate of 16.3 steps
per day .
incurred at the Computation Laboratory , Naval
Weapons Laboratory, Dahlgren , Virginia . ( 2 ) Future Cost Requirements . On the basis of
an assortment of approximately 60 runs per month ,
For the PERT computation system , approximately the running costs at NORC are estimated to be
$ 20,645 of the total of $36,095 was spent on pro $ 5,500 / month . The estimated machine time for
graming, coding , and program checkout . Ten dif these runs is 25 to 28 hours per month, Prelim
ferent programs or computer runs were prepared . inary estimates of running time rates are shown
The total cost is apportioned over 3,600 program in Exhibit XX , on page 84. These estimates are

79
 CHART
FLOW
DATA
SYSTEM
CONTROL
-12
SP
NORC

ASSEMBLE
DATA
INPUT

-12
SP
TRANS EVENTS
LOG
AND
PREPARE
LABEL
TAPE CONTROL
CONTROL CONTROL
DATA REGISTER
TOTALS TOTALS 1
DAHLGREN
TO
SEND
-
-

MATCH VERIFY MATCH

I
|
CONTROL /BATCH
KEYPUNCH
TOTALS CARDS
TOTAL

/BATCH
VERIFY
KEY

80
MATCH

RUN
CONVERT
CARDS
TAPE
TO NORC

RUN
CONVERT
CONTROL 02
e,mb

HHH
TOTALS TO

.
TRANS
TOTAL

RUN
06
DATE
UP TAPE
LABEL

RUN
07
SE
QUENCE LABEL
TAPE CONTROL
EVENTS
TOTAL TOT
ALS

MATCH

RUN
08
COMPUTATION CONTROL
EVENTS
TOTAL LABEL
TAPE
TOTALS

EXHIBIT
XVI
 AND
-12
SP
DISPATCH
DATA
NORC
SHEET
CONTROL

TRANS
AL
TOT EVENTS
IN
CHANGES EVENTS
TOTAL ACTIVITIES
TOTAL
DATE BATCH REMARKS
-12
SP DELETE COMPLETE -1SP2
NORC )( ORC
N NEW NET NORC .
PRED .
SUC

18
XVII
EXHIBIT
 SP- 12 NORC TAPE REGISTER

TAPE LABEL
DATE REMARKS
06 07 08

EXHIBIT XVIII

based on very early figures from checkout opera A complete PERT system will require a control
tions and are not completely accurate . Over the and dispatch clerk to perform the routine functions
first three months of operations , the operating time of maintaining log books and directing the flow of
should be monitored closely and more accurate data .
computations prepared . The dispatch control clerk will be responsible
for gathering all input sheets together , scanning
On the basis of the continuing needs of the them for accuracy , and originating the batch counts .
PERT program , additional programing and follow -up It will be his duty to prepare the transmittal enve
effort at Dahlgren will be required . Two man -months lope and to log the data as transmitted to Dahlgren .
will be needed during November and December . Dur The control data and the output sheets will be re
ing the remainder of the project , support should turned to the control dispatch clerk , and it will be
continue at the rate of one man -week per month . his duty to log in the output and prove the controls
These personnel costs are in addition to the ma before releasing the reports to the designated
chine costs .
people .
A small quantity of the preprinted PERT report
form was supplied to Dahlgren. Continuing effort ( 2 ) Training of Computer Personnel. Five mem
should require on the average of 2,000 sets of this bers of the Sp 12 professional staff were given the
form each month beginning January 1 . “ IBM Aptitude Test for EDPM Programers " (55
234TB ) . The scores are listed below .
2. SP ( COMPUTER ) PERSONNEL
68 )
( 1 ) Personnel Responsibilities . The staff 63 ) A
needed at SP to efficiently perform the computer 62 )
liaison function should consist of two people , once 56 ) B
49 ) с
a full load of events is being processed in the sys
tem . There is a need for a computer liaison officer
and a general clerical helper . The computer liaison The staff member previously selected for the posi
officer should be a professional Sp 12 staff man who tion of computer liaison scored the highest grade .
is fully versed in the PERT methodology. He must However , there is no reason to expect that the first
have a detailed knowledge of the individual com- four people could not be successful in computer
puter programs . The computer liaison officer should work . The person selected began with the PERT
have sufficient detailed knowledge to prepare team August 10 and was briefed extensively on the
specifications for new programs or revisions to PERT system and also on the computer work as it
existing programs . In addition , he should be able was under way at Dahlgren . Several tutoring ses
to verify that such new or revised programs meet sions were conducted on the computer system and
the required or modified specifications . These programs penetrating well into the logic of each
responsibilities fit the flow and control of data to computer run . The computer liaison trainee attended
and from the computer group at NORC as previously a NORC computer programing school at Dahlgren and
described . participated as a regular computer programer trainee

82
 COST NPG - JULY 1 - OCTOBER 31
- -

NORC Labor
Time and O.H. Total

$21 , 740 $ 14 , 355 $ 36 , 095

Approximate Distribution/COST CENTER

NORC TIME

$
Production ( Tables) $ 8,000
Debugging (Tables and PERT) 13 , 740
21 , 740

LABOR AND O. H.
Tables $ 4, 015
PERT 10 , 340
14 , 355

$ 36 , 095

Approximate Distribution / PROBLEM

TABLES
NORC Production $ 8,000
Labor and O.H. 4 , 015
NORC Checkout ( 1/4) 3, 435
$ 15 , 450

PERT
NORC Checkout ( 3/4) $ 10 , 305
Labor and O.H. 10 , 340
20 , 645

$ 36 , 095

EXHIBIT XIX

83
 PRELIMINARY ESTIMATES OF RUNNING TIME

Time
Run Disc Factor Standard (Min . ) Setup

01 CTC Cards 100 Cards (Trans ) 1.0 2.0

02 A , M , B Conversion Trans 100 Trans 1. 3 1.0 (1)

03 Pred Sort Trans 100 Trans .5 1.0 (1)
04 Suc Sort Trans 100 Trans .5 1.0(1)
( 1)
05 Merge Trans 100 Trans .5 1.0

( 2)
06 Update File File 1,000 Events 2.0 1.0

07 Sequency File 1,000 Events 9.0 1.0 (2)

08 Computing File 1,000 Events 4,0 1.0 (2)

09 Sorting Output File 1,000 Events 2,0 20(3)
1.0 (3)
10

Edit Output File 1,000 Events 1, 0

Tape to Card File 1,000 Cards 10.0 2.0

407 Printing File 1,000 Cards 10. O 2,0

( 1)
Setup time for four runs could be reduced to a total of 10 minutes if programs combined.
(2)
Setup time for three runs could be reduced to a total of 1. 0 minutes if programs combined,
(3)
Setup time for two runs could be reduced to a total of 1. 0 minutes if programs combined.

EXHIBIT XX

84
at Dahlgren . As a part of his general training , he its staff in the computer operations so as to allow
has participated in several of the programing sufficient backup in the operations over the next
progress conferences held at Dahlgren . It is desir two to three years .
able that Sp 12 select and train a second member of
VI . PRODUCTION OF TABLES ON NORC
The analysis computation that has been pro 2. TABLE B * , THE TABLE FOR THE COR
gramed on NORC accepts the longest time path to RELATION COEFFICIENT BETWEEN A THIRD
an event as the governing restraint on its accom RANDOM VARIABLE AND THE GREATER OF
plishment. Use of this procedure introduces a bias TWO RANDOM VARIABLES . Table B * was de
into the computation . Theoretical development has veloped in a manner similar to Table A , and the
produced a technique for an unbiased approach . The derivation of the formulae is covered in Appendix C.
NORC has been used in generating tables that will In this case , the functions are normalized so as to
uncover the extent of the bias in the present com make means and variances of two of the three vari
putations . This chapter will discuss the role of ables zero and unity respectively . Thus a nine
NORC in the production of the tables . ( The tables parameter problem is effectively reduced to a five
themselves will appear in a separate report . ) parameter problem with :
The derivation of the formulae for the expected
value , variance , and correlation coefficient of the H – The mean of remaining unnormalized variable .
greatest of a finite set of random variables are de o The standard deviation of the remaining
veloped and presented in Appendix C. unnormalized variable .
The mathematical tables were developed for two p - The correlation coefficient between the first
separate situations . pair of variables .
1. TABLE A , THE GREATER OF TWO RANDOM R- The correlation coefficient between the
VARIABLES . As shown in the derivation , the func second pair of variables .
tions used to relate the two variables under con S – The correlation coefficient between the third
sideration are normalized to fix the mean of one of pair of variables .
the variables at zero and its standard deviation at
unity . The expected value and variance of the The range of parameters used for Table B * is
greater of two random variables then have the the same as those used for Table A. In both cases
parameters: there is a series of formulae for a group of special
H – The mean of unnormalized random variable . cases . The tabular computation for these special
o The standard deviation of the unnormalized cases consists of an evaluation of a known experi
random variable . mental function . The more general case for both
р The correlation coefficient between the Table A and Table B * requires numerical integra
random variables under consideration . tion of nontrivial functions .

The range of the parameters for the tables was se 3. ACCURACY OF TABLES . The general accu
lected to be as follows : racy consideration for the tables was set at four
decimal places . This decision involved a balance
H --5.0 to +5.0 ( intervals of 0.1 ) of both the cost of computation and the potential use
0 0 to 1.0 ( intervals of 0.2 ) of the computed tables . The accuracy dictates the
P -1.0 to +1.0 ( intervals of 0.5 ) precision of numerical integration . The production
runs of the numerical integration used Simpson's
The parameters were selected so as to examine the method . Before the production runs were made ,
function in sufficient detail to identify its critical however , separate test integrations were made
regions . The potential usage of the tables would using the Runge -Kulta method , a longer and slower
not be a tabular form , but in the form of regression method which generally yields more accuracy . For
equations fitted to the tables . The detail of com the test cases , both methods match to 0.00005 .
putations is sufficient for general usage of the
tables and for the development of regression The integration was carried out , starting at an
equations . interval for x , the variable of the integration , of
Table A not only presents the expected value 0.5 . The next step was to divide the interval in
and standard deviation but also the measures of half and to carry out the integration again . This
skewness and kurtosis which are calculated func process was continued until the results of the
tions of the third and fourth moments about the integration changed less than .00005 . At that point
inean . These extra moments were calculated in the last integration was accepted and used in later
order to examine the normality of the functions in calculations for the final determination of the
volved . ( This is important to later usage . ) It can tabular value .
be reported that , except for a few unusual cases , The functions used changed markedly for various
the functions being considered closely approximate values of the parameters . It was , therefore , neces
normality as indicated by the coefficients of sary to compute the range of integration before the
kurtosis and skewness . computing . Several formulae were tested and finally

85
a simple set of rules were devised . The range of and hand calculations at NORC , additional hand
integration was selected to be the greatest of either calculations were performed at SP verifying 18
H + 5.0 or + 7.0 points in Table A. For Table B * , 9 points were
and the least of either verified by hand calculation . All points were veri
H - 5.0 or - 7.0 fied to 4 decimal places .
In addition to checking the tables with machine
VII . PROGRAM MODIFICATIONS
Any design program which is to meet a scheduled 2 to 50. If these estimates are correct , the PERT
date for production must have its specifications system at Dahlgren will consist of small tape files
frozen and made inviolable at some point in its with weekly or biweekly updating runs on each tape
development. The specifications for the PERT sys- of only 2 to 50 transactions . Techniques can be
tem programs were frozen early in August . Since developed to organize these small files so that up
that time only minor coding changes have been dating will be accomplished without 150 to 300
made . separate machine setups . Control systems , to
However , a number of important changes have assure accurate operations , must be maintained for
been analyzed for their impact on the improvement each of the separate component system files .
of the overall operating efficiency of the program
and the analysis . These changes are discussed in The actual increase in computer time will be a
factor of 1.7 times that involved in the integrated
succeeding sections of this chapter . They are
listed in the order of their relative importance . computer run . The following program changes
should be made if component runs are regularly
inade .
1. INDIVIDUAL COMPONENT ANALYSES . At the
time that the design specifications were frozen , only
( 2 ) Data Flow . Exhibit XXI , following this
integrated analysis runs were contemplated . It has
since become ad visable to alter the original plan so
page , shows the modified computer system flow
as to provide for both component and system anal
chart expanded to show only a representative sec
yses . The programs presented in the earlier chap tion of the entire component subsystems . The oper
ters can evolve component analysis . However , in ations for each component system are processed as
indicated in Exhibit II .
so doing they require tight controls and introduce a
strong possibility of human error .
In order to minimize the possibility of errors , it The sequenced component systems tapes (result
is advisable to alter the original program so as to ing from Run 07 ) are merged and combined on Run
11. Each combined subsystem is then reprocessed
accommodate the component analysis without using beginning with computation run ( Run 08) . At this
" ad hoc " procedures . The necessary changes have
point it may be desirable to suppress certain out
been studied and subsequently detailed as follows : puts . This can be done by controlling the param
eters of the sort and edit operations ( Run 09 and 10 ) .
( 1 ) General Specifications
The input tapes to the FBM system merge
- Analyze each component system within a subsys- combine run must be sequenced subsystem tapes .
tem independently and compute all outputs based on After the subsystem tapes are merged and combined
component system only . For example , analyze onto one FBM system tape , the outputs are com
ballistic shell , propulsion , etc. each by itself. The puted on Run 08. The actual number and type of
primary concern at this point is to show " zero " outputs can be chosen , depending upon the selec
slack paths for each component system . tion of the parameters of Runs 09 and 10 .
-Analyze each subsystem independently and com
pute all outputs based on the subsystem only . For ( 3 ) Accuracy Control . Separate control systems ,
example , analyze the missile subsystem and the i.e. , separate control books , must be maintained for
navigation subsystem and others showing the re- each component system . Both control registers
lationship only within each subsystem . Here , the must be maintained for each component system . The
primary concern is to show the “ zero " slack path tape label register must be modified to carry the
for each subsystem . component identification number ( the first three
-Analyze the overall FBM system and compute all digits of the event code ) in the tape label . In the
outputs based on the entire FBM system , as proposed same sense , transaction batches must be set up
in ear lier reports . separately by component system , and in no case can
The proposed method of analysis requires that batches of transactions be combined in the updating
each component system be maintained on a separate operation .
tape reel . It will be necessary to separate all The analysis and correction of errors in the sys
transactions into batches by component system . tem will be much faster and easier , as the recon
This will call for small batches in the establishing struction of any one piece will require the analysis
and updating and maintaining each component file . of a smaller group of transactions and events .
Component files will range from 30 to 500 events , Nevertheless , the maintenance of the controls for a
with transaction batches probably varying from size larger group of files will require more effort .

86
 CHART
FLOW
SYSTEM
MODIFIED

271
209 272 274 261 262 263
] COMPONENT
TAPES
SYSTEM
)(SEQUENCED

11 11
MERGE MERGE
TAPES -
TAPES
LIKE
COMBINE LIKE
COMBINE
EVENTS EVENTS

}
08
RUN
TO
FOR
OUTPUT 27 27
:]
08
RUN
TO
FOR
OUTPUT
. 26
] DUPLICATE
S
SUB
- YSTEM
COMPUTATION COMPUTATION TAPES
)(SEQUENCED

87
11
A
-
MERGE
-
TAPES
LIKE
COMBINE
EVENTS

}08
RUN
TO
OUTPUT
FOR FBM
]
TAPE
FBM
SYSTEM
.
COMPUTATION S
)( EQUENCED

XXI
EXHIBIT
 ( 4 ) Modification of Present Programs . All pro- and the program on Tape 8. Tape 1 is transferred
grams must be modified to verify and print a tape to the output tape ( Tape 7 ) until a key event is
label with a component system code as follows : sensed . Tape 2 is then transferred to the output
Component tape until a key event is sensed . The process is
Problem Run Sequence System Code continued until a key event is sensed on each of
Number Number Date Number Number the six input tapes . The key event on Tape 1 is
XX XX XX- XX- XX XX XXXX compared with the key event on Tape 2. If they
are equal , the predecessors and successors of the
As shown , the component system code replaces the key event on Tape 2 are added to the key event on
week identification . Tape 1. The process is repeated until all the prede
This change will necessitate a slight change in cessors and successors have been added to the key
the computation program . The week number ( the event of Tape 1 .
identification of week " now " ) must be made a part At this point , the key event of Tape 1 is read out
of the setup operation . The program must be modi- to the output tape . If the key events on any of the
fied to pick up this item from the new location . Tapes 2 through 6 are not equal to the key event on
Tape 1 , they are flagged . The program now con
( 5 ) New Runs To Be Programed tinues the process of reading the unflagged input
tapes to the output tapes until the next key event
-Run 11 - Component System Merge -Combine . This on each tape is sensed . At this point , the program
program will merge unlike events from a group of again tests the key event on Tape 1 with those on
component system tapes and combine identical Tapes 2 through 6. Before the test is made, all
events ( key events ) from all tapes . As events are flags on key events are removed . They are only re
combined , all the combined predecessors and succes- established if indicated by succeeding comparison
sors will be added in their correct positions within operations . Identical events are combined by adding
the event message on the combined file tape . The their successors and predecessors to the tape . The
resulting output tape will be in sequence and thus program continues through the merge-combine cycles
ready for the output computing run ( Run 08 ) . The until all input tapes are appropriately transferred
run itself will not produce the duplicate tapes
needed to continue both the output computing and 2. INPUT CONVERSION ( RUN 02 ) . The earlier
the subsystem merge -combine run . Thus , it will be section that describes the input form and procedure
necessary to duplicate the output tape from Run 11 requires that the date of completion and the sched
in order to have input tapes for Run 11 - A, ( see uled date (Ts ) be coded as a number of elapsed
Exhibit XXI ) . weeks . It has been suggested that the accuracy of
( 6 ) Input Restrictions . Event codes for key the input would be improved if this coding were
events , common events among components or sub-
1 done internally . To make this change , Run 02 must
systems , must have at least one digit in the event be modified to perform an arithmetic conversion on
code which is unique to these key events such as the normal calendar date to develop the elapsed
0200000 - XXX . time in weeks and tenths of weeks , The procedure
Input tapes must be sequenced , i.e. , output would be as follows : Pick up the date to be con
tapes from either Run 07 or 08 . verted from the input message . Extract the month
The program can accept no more than six com- portion from the calendar date and subtract 7 ( July )
ponent systems in any one pass , so if more than six from it . Multiply the resulting number by 4:33 to
component systems are to be merged the following develop the first factor in the conversion and add it
procedures must be used . First , select six compo- to a counter ( Counter A). Extract the day portion
nent systems and merge-combine them . Take the from the calendar date and subtract 12 ( for July 12 )
output of this run , mount it with remaining compo- from it . Multiply the resulting number by 1/7 and
nent systems and merge-combine them . Then dupli- add it to the Counter A. Finally extract the year
cate the final output . This restriction is a function portion of the calendar date under consideration and
of the eight tape units of the NORC computer , one subtract 58 ( for July 12 , 1958 ) from it. Multiply the
of which must be reserved for the program tape and resulting number by 52 and add it to Counter A. The
a second for the output tape . Note : Under certain week code read out of Counter A will be equal to the
operating and setup conditions , it is possible to use elapsed time from 12/7/58 to the date under con
the program tape unit for either output or input . sideration . This can be added as a subroutine to
The difference between Run 11 and Run 11 - A the input conversion run ( Run 02) .
( see Exhibit XXI) , is a minor one . The programs
are identical except that the set up operation for the 3. SHORTEST PATH COMPUTATION . It has been
component system merge-combine run includes con- recognized that in some uses it may be desirable to
trol data and instructions . These check the input consider the shortest rather than the longest path.
tapes to verify that they are for the same subsystem . However, there had been little or no field experi
This is not necessary for the subsystem merge- ence to indicate the extent to which this is neces
combine runs , Run 11 - A . sary . The necessity arises when several concur
( 7 ) Mode of Operation (New Program ) (Run 11 ). rent development activities are going on and the
Six component systems are mounted on Tape units 1 plan is to accept the one which meets specifica
through 6. The output tape is mounted on Tape 7 tions in the shortest elapsed time . The problem is

88
confounded by the time restraints of a decision probability calculation would also be inserted in
date . I.E. , a decision will be made at a certain Word 4. If Ts is greater than Tl , and Tl is not
point in time regardless of whether any development equal to T E , a new Ts is generated equal to TL
is finished . plus 1/2 ( TL - Te ) . At this point the probability
This situation can be included in the present calculations are carried out and the result stored in
system by making the following changes . First , the memory . Then an iteration loop would calculate
event record must be changed to allow the use of a several values greater than Is and less than Ts .
special indicator or flag . This flag will be attached After completing the iteration loop , the Ts yielding
at the confluence of the several parallel develop the largest probability value would be written out on
ments wherein a lesser of the restraints type of the output message in Word 4 , along with the new
decision is to be made . Once flagged , the calcula probability calculation . The program would then
tion proceeds as usual , except that instead of using proceed to the next event .
the logic which selects the greatest of several
paths , the program branches and will select the 5. EDIT FOR MISCELLANEOUS OUTPUT . Sev
least of the several paths from preceding events . eral of the runs produce monitoring or control out
After selecting the least of the several restraints puts on the on-line printer . At the present time this
leading to the event , the program should now extract output is being prepared using the standard NORC
the adjusted Ts value . This is compared with the five or six words per line unedited form . Output in
selected least path and the greater of the two , Ts this form is readily understood by only those famil
or Te , is chosen . At this point , the program will iar with the NORC system . If interest grows at SP
continue in the normal manner until it reaches in these intermediate outputs ( such as control totals
another event flagged with the special calculation and completed events ) they should be edited in their
code . respective routines .

4. A PROCEDURE FOR RESCHEDULING . The 6. MORE EFFICIENT SORTING ROUTINES. The

procedure for rescheduling events described in the present sorting runs , Runs 03 , 04 , and 09 , use
Phase I PERT report is quite capable of being strings -of -two sorting methods . There are several
mechanized and can be done as follows . First , the improvements in this method generally classified
probability calculation routine in the computation under the term " progressive " sorting . Such sorting
run ( Run 08) should be added to the edit run ( Run takes advantage of the order which already might
10 ) . Then the following rescheduling subroutine exist in the data . Strings-of-two sorting ignores
should be added to the edit routine . Both of these this advantage . If the future the sorting load builds
routines should be added on an option switch . With up to a point where it absorbs more than 15% of the
these routines , any computed , sorted , or sequenced operating time , improvements in the sorting runs
tape can be used in the rescheduling procedure . would be in order. Exploration of " progressive "
This will allow experimentation at any time with sorting and strings-of-four sorting techniques would
any set of data and will immediately print the re be suggested .
vised experimental schedule without disturbing the
existing file data. If it is desired to add the revised
scheduled dates to the file , the output cards can be 7. GENERAL NOTES . Program revision is a
reproduced into input format on the IBM 519 card necessary part of the continuing data-processing
reproducer . The revisions in schedule can then be operation on any computer . There must be a con
processed as regular transactions using the Trans tinuing review of methods and techniques . Over a
action Code 5 , static change . period of time it is possible to reduce the overall
The program would first compare TL , Te , and operating time by as much as 50% by review and
Ts . If Ts is less than Tl , a new Ts equal to TL
.
adjustment . The costs of such program can be
is inserted on the output message in Word 4. The recovered by increased operating efficiency.

89
 APPENDIX C

SUPPORTING MATHEMATICS

91
 |

FAN
 THE GREATEST OF A FINITE SET OF RANDOM VARIABLES

I. INTRODUCTION

Given random variables X1 , x , together with their joint probability distribution , we would like to
construct the distribution of max ( x1 , xn) , the greatest of the n variables . However , we restrict the
analysis to an approximate solution of the following special case .
We assume that a subset of x1 , ... , X , have a normal joint probability distribution . The complement
1

of this subset consists of variables each of which is constant with probability one .
Let x and y be any pair of X1 , ... , Xn : We shall compute the first four moments of max ( x , y) . If t is a
► .

third variable , we shall compute the coefficient of correlation between t and max ( x , y ).
These results permit the following approximate solution of the original problem . The expected value
and variance of max ( X1 , X2 ) are computed . Assuming that the distribution of max ( x1 , x2 ) can be adequately
approximated by the normal distribution, we approximate the coefficient of correlation between X3 and max
( X2 , X2 ) . Still using a normal approximation , we approximate the expected value and variance of max (X3 ,
max ( X1 , X2 )] which is max ( X1 , X2 , X3 ) . After a number of steps we reach max ( X1 , X2 , ... , xn) .
The third and fourth moments of max ( x , y) are not used in the above computation . These moments give
some insight into the accuracy of the normal approximation .

II . NORMALIZATION

We set aside the trivial case in which the variances of both X , and X2 are zero . We apply a linear
transformation ax ; + b simultaneously to X , and x2 to obtain the transformed variables x and y with E( x) 0,
o 51 , and p (x , y) = r, where E, V , and p denote expected value , var
V ( x ) = 1 , E (Y) = M , V (Y ) = 02 , 03 02 1

iance and linear correlation coefficient, respectively .

As we consider pP [ X3 , max ( x , y )] , we note that p is undefined if V ( X3 ) = 0. This special case is set
.
=

aside. Then a linear transformation reduces X3 to t such that E (t) = 0 , V (t) = 1 , p (t, x) = R , and p (t , y)
= S. Since our interest at this point is in p [X3 , max (X1 , X2 )] , the linear transformation applied to X3 is dif
ferent from the transforma tion applied simultaneously to x , and X2 :

III . TABULATION OF RESULTS

Tables have been constructed from the following formulas . They are in the possession of SP . A sepa
rate report will be issued in which the use of these results will be demonstrated . Although the following
formulas are expressed in moment notation , their subsequent utilization should be in the form of expected
values , standard deviations , coefficients of skewness and coefficients of excess (kurtosis ).

LIST OF RESULTS :

Let Vi i = 1 , .
4 be the moments about zero of max ( x , y ) .
.

CASE I. If 0 < o < 1, -1 < r < 1 , then

X - H X -
rx X -r
x2 o
1 1 X -M

ans ( )- "). o
V

Vi
277
xi exp
2
F

1 - p2
+ exp
[*** ]
2 0
dx ,
93
where
х

1 t2
F (x) :

- 1 у 2п

CASE II . If o < o < 1 , г = + 1 , but not simultaneously o = г = 1 , then

— ОЮ
exp
(S) 2
dt .

1 + о 1 2

Р. = p F
(:). о 2п „ C:( :))
exp
2 1 to

в ( 1 + с) 1 2

2
v , = 1 + ( и 2 + O2 -1 )
(:) « (E:( :))
F
1
+

у2п
exp
2 1 +

2
р и ? ( 1 + o) + 2 ( 1 + оз
3 ( из + 3 ро ? ) F

(, ) о:( :)) +

V 2п
exp
2 1 +

2 1
3 + ( p4 + Cu ? о ? + 3o4 – 3 ) F [ *3 ( 1 + 0) + (3 + 3o + Зо ? + 5о? ))
:-)
+
2п

2
1
exp
[:(:)) 2

=
CASE III. If o = r = 1 , p < 0 , thеn vi = 0 , v2 = 1 , 3 = 0 , v4 = 3 .
CASE IV. If o = г = 1 , p > 0 , then

V1 = , V2 - 1 + 2, 3 = 3р + 3, 4 = 3 + 6u? + *.

CASE V. If o = 0 , then

1 exp
vi = p F (и ) +
и ( :)
уп 2

1
v, 1 + ( ? - 1 ) F ( и) +
у 2п
и exp
(:) 2
2

v 3, = F (р) +
.. 2п
( и ? + 2 ) exp
" (с) 2

2
1
3+ (и “ – 3) F (и) +
-

V 2п
( 3 + 3 р) exp
(c ) 2

94
CASE VI. If 0 < o < 1 , r £ + 1 , then with the notation R = p ( t , x) and S p (t , y ) we have

X -μ
rx

)---)。
(三)
R o
E [t max ( x , y )] x2 exp F dx
27 S 2
Vi -p2

X -H
S x ( x - 4) 1 х
f X - r o
+

271
OV s
00
exp
[ *(*6=-)] F

V1 - P2
p2
dx

+ H ( 1 - 10) (rS - R + TOR - OS) к?

V 271 ( 1 - 250 + 02 ) 3/2 [

exp
., 2 ( 1 – 2ro + 02 )

CASE VII. If 0 < o < 1 , r = + 1 , but not simultaneously o = r 1 , then

R[1-6 7692)
E [t max ( x , y )] = R1 - (1 + 0) F

CASE VIII . If o = r = 1 , then E (t max ( x , y) ] = R.

1

CASE IX . If o = 0 , then E (t max ( x , y )] = R ( 1 - F (u)] .

INTEGRALS REQUIRED IN THE DERIVATIONS.

Well-known properties of the normal distribution give :

( x - a) 2
dx = by2
b 27
( [ 43+
exp
2b2

(x - a )2 (x -a) 2
Š x x0[
-00
X exp
2 b2 ] -S
dx [(x -a) + a] exp
[ 2 b2
dx = 0 + ab V211 = ab 27 ,

(x - a ) 2 a

-00 -C (****) dx = 0 var

exp
2 b2 (*
F b

с
( x --a) 2 ( x - a)? ( C -a )
x exp dx [( x – a) + a) exp dx -b2 exp

(- 2b2
) [. ) --[ 3
2 b2 2 b2

с a

+ ab 25 (**) F
b

95
If we use the notation

a
1
xi exp dx ,

Van s uses ( . 2

1
xi exp
J
271
a
( ) 2
dx ,

we compute using integration by parts

L =

271
exp
(3 )
2

1 a2
12
271
a exp
( *3 ). 2
+ F ( a) ,

1 a?

ma (+) 277
( a? + 2) exp
2
1

1
( a2 + 3a) exp + 3 F (a) ,
1
va +36) :
2п 2

1 a2
J, =

12 п4 .
Vevo (**) exp
2

1 a2
J2
2п
a exp
( *).2
+ 1 - F (a) ,
van

1 2
J3
2.
(a ? + 2 ) exp
( ) 2

1 a?
JA
271
(a 3 + 3a) exp
(*) - 2
+3 3 F ( a).

DERIVATIONS .

CASE I. The probability density of x and y is :

1 1 у – и
f ( x, y)
270V1 - ?
exp
{ 2 ( 1-12 ) [13-2- Y . (*::) ]}
2rx

96
which can be reduced to the following two forms

1 1 у
f ( x , y) =

20071
exp
[ * )]• [-zd -ris ( -- :-)']
2
exp
2 (1 )

1 ? 1 2
exp
f ( x, y)
2 πσ1 - 12 ( 3 ) [ 20-7767_ )']
2
exp
2 ( )

We have
y

Vis
SS (max ( x , y)] ' f ( x , y ) dx dy. A

We shall use notation whose meaning is indicated in the figure .

- X

Ş . SS
+

We note that max ( x , y ) y in region A and max ( x , y ) x in

region B.

oo у
1 1 1

SSA 2πσ1- 12 - ( ( :-) ]-- [(

y ' exp
2
у
dy exp
2 ( 1 - p2 ) (- 2 :))]]
X -
у
dx

y -M
1 у y - r O

cookieſ" en [11(95%)]var
2πσν * ] vir у exp
2
1 - p2
dy

- H.
1 у 0

] -27 ). y exp
--
1 -- [ )1-6
O V271

oo
2

x2
F

х
p?
dy .

(*=*_-)'] .
1 1

SS
B
2πσι --( xi exp
2
) dx exp
[ 2 ( 1 - pl )
dx

he
1 o -
rx
x exp

完了 … ( ) ( ) 。 *2
F dx .

The result for Case I given above is obtained directly .

97
 у – р
CASE II . Since r = + 1 , + x , y = p + ox .

о
р
х > y = > x > p tox , x (1 + 0 ) > px > 1 + 0

The last relation has meaning because we do not have o = r = 1 .

р
y = p + ox , if x < 1

1 to
Max ( x , y) =

Hence
... .
x , if x >
р
1 + 0

р
1 + 0
2
1 1 х

т.
| (и + ox) exp
(:) « = | dx + xi exp
( 1) dx .
II

2
-

2п. 2п
-||

Е.
1 + 0

Using integrals listed above , we write

р 1 р

• r(:):-) [-Е ( :.) ]! " [ :([ ::-) ]

, - д F
+
* о
2п
exp
+ 0
+

2п
exp
1

1 о
д F
- " ( г.). " [:(:)))
+
+

2п
exp
2 1 +

р -1 р
2 и? F
{4
P ( 7.) +2-0 Х. [ (7.) ]
+ 2п
exp
2 1 + .

-1
+ O2
2п : " [:(:)1) - (, :)
1 + 0
exp
2 1 +
F
+

1 р р
+

" 21 1 + о " [:(:))

, -- :-)
2 1 +
+ 1 F
1 +

р ( + 0)
= 1 + (р ? + а2 – 1 ) F
(= ). » [:(:))
о 2п
exp
+ о

98
 -1

• ' ( : )... | ""[ :(:-)

Р. = г
(-18
1
+ Зu?
2п
exp

-1 р 1

..." ,
+
V2п 1 + о " [:(G:)) -:( :-)
exp
2 1 +

и?
" { [ ] - [ :( :-) ])
+
2п L1 # о) ?
2.
+ 2 exp
1 +

1 ? 1

L. ) [:(.:..)))
( 2п L( 1 + а) 2
+ 2 exp
2

2
р р ? ( 1 + 0) + 2 ( 1 + o3 )
(из + 3 ро2 ) F
(, ).
1 F
+

2п » [:(1
exp
:-) 1 +

2
-1 1

v, = 1 г
"(1ta)- •• 4 + о
+ 4
2п
exp

[:():-))} 2 1

-1 р 1

**
+ 6 p ?o2
2п
1 + 0
exp
[ 2 1 F
Ә)1 ( :-)
о
+ F

-1
и? 1 р

... ... ...) [:()

+ 4 доз 3

- 11} 2п (1 )
+ 21 exp
2 + а

-1 1
+

" , [.." ..] » [:(:-)) -- (- )

... 2. L( 1 + )3 1 +
exp
2 + о
+ 3 F
0

3 Зр 1

2п [.. ... ) » [ :(- )1.5-2 г( :-)

( 1 + 0) 3 1 + 0
exp
2
—

1 + 0
+ 3 3 F
1 +

р
3) F
3 + (и + + € p ? o2 + 3 o4
- ( -) 1 + 0
2

+ р * ( 1 + 0) + (3 + 3 + 3O2 + 503 ) 1

у 21
exp
[:(:)) 2 0

CASE III . If o = r = 1 and 4 < 0 , then clearly max ( x , y ) = X. Since x is a standard normal variable ,
=

= 0, р , 1 , V3 0 , P4 3. These moments as well as the moments given in Case IV are well known
-

parameters of the normal distribution .

99
CASE IV . If o = 1 1 , and x > 0 , then clearly max ( x , y ) y and vi = H, V2 = 42 + 1 , vz = H3 + 3H ,
* + 642 + 3 .

CASE V. If o = 0 , then max ( x , y )

M. if x < u
X , if x > Ho

1
v = 4 ' F (u) + xi exp dx .
van s 1000( )
2п
H
2

1 કર

--- (+ )
V = \ F ( ) +
2п
esp

1
v,
2 = 1 + ( h ? - 1 ) F (H) +
van exp ( :). 2

1
K3 F ( u) +
‫ע‬
V3
271
( u + 2 ) exp
( -) 2

5
1
= 3 + (4 * - 3 ) F ( M) +

CASE VI . The covariance matrix of t , x , and y is

271
(u? + 3 р) exp
( ) 2

1 R S

The determinate Al of A is 1 + 2 rRS

Δ .

0 : )
r2
R

R2
1

- Sa. We know that r, R , and S must satisfy A1 > 0 .

1

We assume that the joint distribution of t, x , and y is normal . As indicated in the section on Normalization ,
1

we assume that E ( t) =
E ( x ) = 0 , V ( t)
=
V ( x) = 1 , E (y) = M , and V (Y) = 02. Hence, if |Al > 0 , the
=

joint probability density of t , x , and y is

1
f ( t, x , y) -

(27)3/2 o VTAL
exp
() 2

where

1 у M
φ
( (1 – 12 ) t2 + ( 1 - S? ) x2 + (1 – RR ) (* ) + 2 ( rS – R ) tx
II

141
-
у H y HL
+ 2 (rR S) t + 2 (RS – r ) X
0 o

100
The limiting case with 111 = 0 is discussed below , and we now assume that 101 > 0 .
Since we shall integrate first with respect to t , we reduce $ to

2
1 - p2 rS - R rR - S y 1 у -
$ t rx
191
+

1 - p2
X +
1 - p2 :-):-(1-24 *** 6 --)]
0
+

1
+

Using this expression for we write

1
E (t max ( x , y )]
( 272 )3/2 OV111 STS t max ( x , y ) exp
( 3) ata2
dt dx dy

1 1 y M
max ( x , y ) exp 2 rx
(277)3/2 0 MAI SS oo
{ 2 ( 1 – 12 ) [ 0

y ) 1 r2 rs R R - S y - M

.
dt
t exp
+(---)}]}« ſi
+

S [ dx dy
2 14
t +
1 - 82
X +
1 -
r2 -)]

As indicated in the section above concerning Integrals , this reduces to

oo

1 1 у ML
max ( x , y ) exp x2 – 21 x х

(271)3/2 OVAL SS
3
2 (1 - 12;/ * o
(---)]} *
rS - R rᎡ - s Ꭹ 191
V27 dx dy.
[ -]
1

1 - p2 1 - p2 o - p2

We shall simplify and proceed as in Case I.

oo
S
-
-

Inax ( x , y ) R rs rR V -
E ſt max ( x , y)] X + X
2πσ r2 1 - p2 1 - p? -]
o

1 y - у
exp
{ 2 ( 1 - r2 ) [<(3-2015> * ( "))}««v
] - S.S
rx
0
+

o
dx dy

A
+

101
where the meaning of A and B is indicated in the figure above .
Repeating some manipulations used in Case I , we write ,

-
1 у R rS S - rR y - H

x
SS 2015 »
A
2πσ1
y exp
] }S
»[-+63"))" 2
dy
1 - p2 1 - 12 0 * ). X

2
1 у
exp
[... 2 ( 1 – r2 )
o
1)0 ] dx

1 у R rs у
dy

Sven [14 "))

y exp
=**(--: --)
- ] }[A
11

2 1 r2
2π σν1

R у – S - R y 1 у –
dx

**** -] - [ zu - ) ( -:**"*)] "

+
1 r2
+
1 r2 o
exp
2 ( 1 – 12 )

у
00 у
1 R - rS

2001--
2πσ : -
(:-]-
5»s «[;- r2
y exp
2 1
[- ( 1 - 1 )] exp
21
o

2 ( 1 - 12)
-
r2

У .
y - r
TR - r2S + S - IR y - H -
0
V1 - 12 V27 F dy
1 - r2 0 1 - r2

rS - R 1 у у

2πσν1 - r2 s
-oo
... [ -
y exp
2 ( 1 - 82 )
y2 - 2ry +

( =“)]
o
dy

у H
y - r
s

了 -06-6
y ( y – u) у o

12
0 2 o
exp
[ ( ;--)] F
1 - p2
dy .

х
x2 S - у -

SS
B ( 66* --x)
9 -9
. 2
dx
1
2
0

s - IR R rs 1

1 - p2
rx + -
1 r2 ]: exp
2 d --:)]
[- 2 ( 1 - 52 )
rx dy

102
 2
д

-
rx
1 s TR 0

2πσν1
1.- (5)
x exp
2 1 - 12 -
(- ( 1 - r2 ) o] exp
2 ( 1 - r2 )

-
X
H
rx
rS - p2 R + R - rs 0
+

1 - 12
xoOVI -
r2 27 F
r 2
) dx

TR – S 1 х
- X -
x2 2r x
271 V1 - 52
x exp
{ ad 2 (1 r2 ) ***( --)]} : +
0
dx

R x2 o
\-

***
為了 ( 7)
+

2 NT
x2 exp
2
F
1 r?
). dx .

In the expressions for

«şuan $$me
1 х
X exp
{ 2 ( 1 - 82 ) 6
[1-3(=:)}} - 2r x

which will be evaluated next. The integral reduces to

1 2ro + o? ( ro - 1 ) ?
ſ{ x exp
202 ( 1 – 12 ) [ X +

1 · 200 + o ? "]} exp

[20-20-on]
2 ( 1 - 210 + 0 ? )
dx

H ( 1 - ro)OVI1 - 52 -

2 п
= exp
2 (1 2r 0 + 02 ) ] 2r 0o + 02 V1 - 2r 0 + 02
1 - 20
2 πμσ ( 1 – Γα) 1 - r2 -

( 1 - 21 0 + 02 ) 3/2
exp
[ za -toront]
2 (1 2r + )

We substitute this last expression in SS s

A
+

B
and get

rS - R TR - S --

E [t max ( x , y)] + х
2πσ 1 -- Γ2 277 V 1 r2

103
 V2 π μσ ( 1 - Γα) 1 - r2 к?
(1 - 2r o + 02 ) 3/2
exp
[ za - on]
2 ( 1 - 2r 0 + 02 )

у H
у
S y (y - M) 1 M
+

σ ν2 π ***- [- (-97673 ) exp

-:( : )]
2

X
F dy

le
rx
R x2 0

2 71 s x2 exp

(5)-(4 )2

which reduces immediately to the result stated above for Class VI.
F
r2
dx

The formula for E ſt max ( x , y)) is defined and continuous along 191
.
O for fixed r. Hence , limiting =

values of E (t max ( x , y)] are correctly computed by the above formula . The continuity of E (t max ( x , y)]
along 141 = 0 has not been completely investigated by the author . However , the following special cases are
considered.

First suppose that R + 1. Then 1 + 2 rRS - p2 - R ? - S2 ( r 7 s ) . Since 191 > 0 , we must

have s trand 141 = 0. In this case t = + x and (sincer # + 1 in Case VI) the probability density of x
and y is

1 1 у H у

2710 V1 - 52
exp
2 (1 r2 )
x2 2 rx +

( --)]}
1 1 y - у
dx dy .
Etmax ( x, y)]
2πσ 1 - 1V SI
-00
+ x max ( x , y ) exp
420-75[12-2013**
{ (1 (***)]}or as )
2rx
o
+

In the general discussion above in which 141 > 0 , one finds the relation
co

max ( x , y ) R rs S rᎡ Ꭹ
E ſt max ( x , y )] X + X
SS
-oo
2πσν1 - 12 1 r2 1 r2 -]
1 у 1 y
exp 2 rx

{{ :2 ( 1 – r2 ) [13 Y " .(*:-) ]} 0

+ dx dy .

If in this expression we replace R by + 1 and S by r (as established above ) , we obtain the immediately
preceding expression for E [t max ( x , y)) . Hence , the result when R + 1 is obtained from the general re
sult with R + 1, S + r.
In the same way we can show that the general formula can be used if S = + 1. In this case R + r.

104
CASE VII . First assume that R = # 1. Since r = + 1 , Y - t = + x , y = H to x, and x > y if and
only if x > H + o X or x > Hence ,
1 FO

р
He to x , if x <
1 + σ
max ( x , y)
He
x , if x >
1 F

Since the joint probability density of t and x is

1 1

2 п 1 - R2
exp
[ za! 2 ( 1 - R2 )
( t2 - 2 Rtx + x2 )
2017
1 1 x2

ਜਵਾ ,ਜੇ- Rw*) eu (( )

2 71 V1 R2
exp
2 (1 – R )
( t - Rx)? exp
2

1 x2 ( t - Rx)2
-

Etmax ( x , y )]
277V1 - R2 s ( 1) .s - (... ) o
max ( x , y ) exp dx t exp
2 ( 1 - R2 )
dt

1 x2
Rx V11 -- R2
R2 V27 dx
2п1 -
R2 s.
-0
max ( x , y ) exp
f ) 2

H
1 Fo
R x2 R x2
( u to X ) X exp dx + x2 exp dx .

vis
V 27 -00 -> (47) VI * ( )
2
р

1 F o
2

sing the table of integrals derived above, we reduce this last expression to

2
-1 -1 р 1 le HI
RH
2 п
exp

[ :( +..
.)'] [ :( .) )] " (+-)} RO
2 п 1 F o
exp
1 Fo
+ F
Fo

1 He 1
+ R
2 71 1 Fo
exp
([ (0 7.) ] - (17.)
2
+ 1 - F
1
=

R 1 - (1 F o F
--[ 1-01 70) *(17.)] 1 to

105
 If R = + 1 , then t + x = Rx , and

US
+
1 Fo
1 x2 1
Eſt max ( x , y)] =

2 п
S Rx ( u I ox) exp
(c ) 2
dx +
2 ſ H
Rx2
ex **) dx .

1 Fo

Since this identical expression arose above when R # 1 , the results derived above still hold when
R = t 1.
(The case s + 1 is included in the case R = + 1 because s = + 1 and r = + 1 imply R = + 1 ; in
deed R rS . )
The analysis of Case VII is complete .

CASE VIII . If o = r = 1 , we consider four cases , namely

>

(a) H 30 , R + + 1 ( c) x > 0, R 1
( b) u < 0 , R = + 1 (d)) u > 00,, R = + 1 .=

In case (a) we have max ( x , y) 1 = X.

1 1
Eſt max ( x , y)] =
tx exp (t2 - 2 Rtx dt dx
271V1 - R2 -

‫ار‬ [ zu - 2 ( 1 - R2 )

1 x2 1

2 п 1 - R2 sj ਜ (** } ( [[
xexp
2
dx t exp
2 ( 1 - R2 )
(t Rx)
]• dt

oo

1 x2
xexp Rx V1 - R2 dx

sovino
27 V1 - R2 2
)
-
2 п

R
x2 exp
=

27 00
(( **) dx

= R

because the coefficient of R is the variance of a normal distribution :

Case ( b) . We have t Rx , max ( x , y) = x , and

1
E [t max ( x , y)] Rx2 exp dx R.

van 5s **** (***)

2 п
-
2
-
106
 Case (c) . We have max ( x , y) = y , and
.

oo

1 1 у у
ty exp St dt dy
Eſt max ( x, y)]
2πσ 1 - 32 s
-
{ za [1-28 *** :(---)}
2 ( 1 - S2 .
+

1 1 1 у
=

t exp
2110 V1 - S2
y exp
*]
sven [:(32)) 2
dy

5 +en [20-65(-254
( )']] (1 )
dt

1 y y
=

..ovo Sven [ -
y exp
4 :-)] o
***** 17- si vor o
2s
o
52 2 п dy

oo

‫أنه‬
=

y (y - u) exp
[ 4 ] 2
dy

S 1 у H S
dy

‫۔‬ o² √27
(y
[ 14 )].
----(:4 w [-16
1 ----- 17 :)).]
exp
2
dy +
o
( H ) exp

S
( 02) + 0
11

= R

because s R when r 1.
у H у — р
Case ( d). We have t = + = R max ( x , y) y , and
o

1 у He у
Eſt max (x, y)] = R
OV2
y exp
(* ]«
[ 6=-)
2
dy = R

as above .
The analysis of Case VIII is complete.

107
CASE IX . When o = 0 , we have

H, if x < Meo
max ( x , y) =

x , if x > H.

If R + 1 , E [[t , max (x , y)]

y ] =

oo

1 1
( t2 2Rtx ) dt dx

27V1 - R2 S x]
t max ( x , y ) exp
Ezu -
2 ( 1 - R2 )
-

1 1

2.vier ]
dx
max ( x , y) exp
() Š t exp
[ za -R. * - Ray] de( t - Rx ) dt
11

2 2 ( 1 R2)

10
1 - R2 oo

1 x2
max ( x , y) exp RX
Rx V1 - R2 V2
a T dx
2.vices
2 п 1 R2
(*) 2

р
R x R x2
dx + x2 exp dx

venej 27
M X exp
)
(5) 2
ventas
2 71
р (9 2

?
Ru F ( M)
=

)2 [[...
[wan (**)]--
2 п
exp
1-- P- « ]
-5).---
** exo (* + R
2
+1

( - F (u)]]
= R 1

If R + 1 , then t + x = Rx , and
8

1 x2
Eſt max ( x , y) Rx max ( x , y ) exp dx

是ES
V2.71 -00
( **) 2

oo

R x2 R x2
H X exp dx + x2 exp dx

vajE
2 п
-
((7) 2
ve 1 ( 3 )
2
2

= R

as above .
The analysis of Case IX is complete .

108
# U.S. GOVERNMENT PRINTING OFFICE : 1962 0-656132
7
 WILS QUARTO
1993 .P761 x 1958
PROGRAM EVALUATION
RESEARCH TASK

APR 0 2 1990
 UNIVERSITY OF MINNESOTA
wils
Quarto V 993.P761x 1958

Program Evaluation Research Task ( PERT)

3 1951 002 472 361 L