HP Calculator

HP Solve
Calculating solutions powered by HP
In the Spotlight
Issue 25
» 30th Anniversary Edition of the HP 12c October 2011
Performance never goes out of style.
Celebrate 30 years of this one of a kind Welcome to the twenty-
business calculator with a limited edition fifth edition of the HP
collector's special. Solve newsletter. Learn
calculation concepts, get
advice to help you
Your articles succeed in the office or
the classroom, and be
the first to find out about
new HP calculating
solutions and special
offers.
» Download the
PDF version
» The HP 15c Limited Edition » The HP-12C, 30 Years and
of newsletter articles.
calculator Counting
One of nine machines in the Richard J. Nelson and Gene Wright
Voyager Series of HP » Contact the editor
Read this extensive review of
calculators, HP announces a how the HP-12C calculator From the Editor
re-release of their legendary has withstood the test of
HP 15c calculator. professionals and time
longer than any other
calculator.
» Calculator » Converting Decimal Learn more about

Programmability—How Numbers to Fractions current articles and
Important is it in 2011? Joseph K. Horn
feedback from the latest
Richard J. Nelson
Joseph explains the Solve newsletter
In this article, learn if there is continued Fraction Algorithm including One Minute
an advantage to having a and provides an RPN HP- Marvels and a new
scientific calculator 15C program that quickly column, Calculator
programmable as well as the converts any decimal to a Accuracy.
reasons for programmability. fraction to the accuracy set
Learn more ‣
by the FIX display value.
» Limited Edition HP 15c » How Large is 10^99? » Fundamentals of

Execution Times Richard J. Nelson Applied Math Series #8
Namir Shammas Richard J. Nelson
And how important is this
The HP 15c Limited Edition limit? Three examples of the π is probably the most
is here and comes with the largest possible physical well-known of all
same features of the original values are used to see if mathematical constants.
with added speed. How fast, 10^±99 is big or small This overview provides
you may ask? This article enough to calculate numbers several resources and
answers this question by in the real world. links exploring why π is
comparing the speeds of the such an interesting
old HP-15C and the new HP number.
15c LE.
» Update Profile » Change Email » HP Home » Support & Drivers
HP respects your privacy. If you'd like to discontinue receiving e-mails from HP regarding special offers and information,
please unsubscribe here.
For more information regarding HP's privacy policy or to obtain contact information, please visit our privacy statement or write
to us at: HP Privacy Mailbox, 11445 Compaq Center Drive W., Mailstop 040307, Houston, Texas 77070, ATTN: HP Privacy
Mailbox.
© 2011 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.
30th Anniversary Edition of the HP 12c
HP Solve #25 page 3
From the Editor
Article – Next 
Return to Top
HU U
HP Solve # 25 Page 4 Page 1 of 1
PERFORMANCE NEVER GOES OUT OF Style
HP is celebrating the 30th Anniversary of its iconic gold trimmed financial calculator, the HP 12c.
With its sophisticated look and feel, this calculator is still in its heyday, withstanding the test of time
for professionals in the world of banking, finance, and real estate.
In 1981 HP introduced the 12c---so perfect that

very little about this calculator has changed. It
was the beginning of the mobile computer and
was not only adopted by movers and shakers, but
by everyone! Today, people marvel that an
electronic brought to market 30 years ago
continues to hold value through performance,
functionality and durability while new technology
reaches obsolescence within 3-6 months.
On May 4, 2011 the Wall Street Journal referred to the HP 12c as the “Cult Calculator” asserting that
for finance professionals, “Ignorance of the 12c can flash more warning signs than a scuffed pair of
shoes.” This article became HP’s inspiration for the 12c 30th Anniversary ad campaign which will
run in numerous financial publications word-wide. The ad portrays a well-dressed, sophisticated
business man who has successfully advanced his career from a broker fighting his way through the
bull pen on Wall Street to a talented financier in the boardroom of America’s finest corporations—all
with the assistance of his trusted companion—the 12c. The WSJ article states: “the 12c is just as
commonplace for financial analysts as papers and cell phones” demonstrating that it has become more
than just a tool of the trade; it’s a fundamental staple.
Exclusive to the 30th Anniversary, HP is offering this exemplary calculator in a limited edition
collector’s special. Each 12c comes in a leather gift box, is embossed with a “30th Anniversary
Edition” emblem, and is stamped with a unique production number, making it the perfect gift for
business professionals, students, and calculator connoisseurs.
The 12c “30th Anniversary Edition” is available for purchase September 1, 2011 in retail stores and
online at: www.hp.com/go/voyager
*Peterson, Kristina. “A Cult Calculator Endures.” Wall Street Journal, May 4, 2011: Print
http://online.wsj.com/article/SB10001424052748703841904576257440326458056.html

The HP 15c Limited Edition calculator
HP Solve #25 page 6 HP 82240B IR Printer Introduction
 Previous Article – Next 

U UH HU U
Return to Top
The Legendary 15c----BBaacckk BByy PPooppuullaarr DDeem
ma nd
an d
Introduced to the market in 1982 as a part of the HP

Voyager Series, the 15c scientific calculator quickly
achieved a loyal following as fans marveled at the
calculator’s premier performance and functionality.
After 7 successful years in the market-place, HP
discontinued the 15c and replaced it with multiple line
display calculators.
This decision devastated 15c loyalists. Since then, they have beaten their chests loudly for HP to
bring it back, creating websites such as, “Bring Back the HP15c”.
Their passion for the dearly departed 15c was so strong that over 15,000 people in 175 countries
have signed a petition—“HP, bring back the 15c.”
With demand on this petition exceeding 84,000 units, bidding wars on Ebay reaching over $500,
and tireless forums of fans lamenting the loss of “The long gone but not forgotten” 15c, HP
relented—“It’s time to resurrect the legendary 15c!”
The 15c PR campaign will commence on September 1, 2011 and be featured in numerous
engineering and scientific publications world-wide. The ad is a play on the Voyager Calculator
Series and the Voyager Space Mission— both symbols of the technology revolution during the
early 1980s. The Voyager Space Shuttle discovers the 15c on the outer edges of the Milky Way,
floating further and further away from planet earth and its devotees. The Golden Record,
intended to relay messages from human life to extra terrestrial life, communicates the sentiments
for the return of the 15c.
Today, fans everywhere can celebrate the return of the HP 15c with this limited-edition
collector’s special. Each 15c comes in a leather gift box, is embossed with a “Limited Edition”
emblem, and is stamped with a unique production number, making it the perfect gift for 15c
aficionados!
The 15c Limited-Edition Calculator will be available for purchase September 1, 2011 in retail
stores and online at: www.hp.com/go/voyager.

The HP-12C, 30 Years and Counting

U UH HU U
Return to Top
The HP-12C, 30 Years and Counting
Richard J. Nelson and Gene Wright
Introduction
HP has a long list of calculator firsts(1) and the 1981 HP-12C
business, finance, and real estate calculator significantly
contributes to the list. The 12C has been continuously made
for 30+ years. Made in five countries(2), and often in two
countries at the same time, HP-12Cs made today look pretty
much the same as they did 30+ years ago. The HP-12C
calculator has withstood the test of professionals and time
Fig. 1 – HP-12C 30 Anniversary Edition. longer than any other calculator (and possibly longer than
th
any other consumer electronics product ever made).
The financial HP-12C was introduced September 16th 1981 along with a scientific HP-11C. These two
machines were eventually joined by six other similar looking calculators as part of the Voyager series of
HP calculators. See Fig. 2 below. Also known as the “Slim Line(3) ” calculators they were quite different
from other calculators in HP’s line up because of their smaller size and thinness. Some sellers/reviewers/
websites also refer to the voyager calculators as the “10 Series.”
Fig. 2 – The HP-12C joins eight other voyager series calculators.
The HP-12C integrates the five Time Value of Money, TVM, variables into one solver that provides the
easy and very accurate solution of n, i, PV, PMT, and FV using a consistent approach to solving complex
compound interest problems.

Why is the HP-12C so popular?
The reasons are many, and which ones you will hear or read about will depend on the knowledge,
experience, and background of the reviewer. Here are a few reasons in no particular order.
The HP-12C has a clear readable display: The popularity of the HP-12C was debated at a user’s
conference in London in 2002. The intent of the one hour discussion was to identify the most important
features of the HP-12C that makes it so popular. Six qualities of excellence were given to start the
discussion. Everyone had an opinion, but after an hour one quality was mentioned over all others; a clear
readable display(4). The HP-12C uses a 10 digit 7 segment Liquid Crystal Display, LCD, with 0.2 inch
high digits.
The HP-12C has a clean uncluttered keyboard: The HP-12C is a tool to be consulted in the “heat of the
deal” for the business community. It must not appear intimidating to either the customer or the user. The
gold display bezel and the gold frame around the keys give the calculator a timeless quality look. The
gold is in contrast to the textured dark brown case. Of course the classical HP calculator keyboard key
feel and tactile feedback click is also an important part of the keyboard. Two shift keys, yellow and blue,
are used to provide access to the 120+ functions of the calculator.
The HP-12C uses a horizontal format: Most of the calculators HP makes use a vertical format. The
Voyager series of calculators, however, broke from the tall thick vertical cases of other models to use a
thin horizontal format. This was quickly accepted for two reasons. First, the intended user was often in
the office and at a desk. Second, the light weight, 4.1(5) oz., makes it easy to hold in the hand. Holding a
machine in your two hands and operating the keyboard with your thumbs (pre-cell phone) is natural for a
12C if used while standing.
The HP-12C has a long battery life: The HP-12C is famous

(as are all Voyager models) for the long life of the battery.
The original HP-12C uses three button cells(6) (Eveready A76
alkaline or Eveready 357 silver-oxide) and the current
incarnation uses two coin cells (Lithium CR2032). The
original “Owner’s Handbook and Problem-Solving Guide” Fig. 3 – Button (left) and coin HP-12C cells.
dated February 1981, page 216 says, “In ‘typical’ use, the HP-12C has been designed to operate 6 months
or more on a set of alkaline batteries. The batteries supplied with the calculator are alkaline, but silver-
oxide batteries (which should last twice as long) can also be used.” In today’s world of battery operated
devices this is an unbelievable understatement. Few battery operated devices run as long as the
manufacturer says. Users report multiple (3-5) year battery life.
The HP-12C has a powerful business function set: The HP-12C has a 2,465 year (1582 – 4046) calendar
built in with a nice set of date functions. It also has an excellent mix of business and statistical functions
with the normal [no trigonometry(7)] math functions. Specific applications such as IRR and NPV are also
included. See Appendix A for the details.
The HP-12C uses RPN: Most business calculator users have no idea what RPN means, but they easily
learn the beauty of simply pressing a key to execute a function. You want to add two numbers you press

the + key. You could have divided as well because you give the machine the number(s) and then you tell
it what to do.
The HP-12C is programmable: See article on the importance of calculator programmability elsewhere in
this issue. As with RPN most business users of the HP-12C are not very concerned with its program-
mability. When they have a number of similar problems to solve, however, they will discover how RPN
programmability may be easily learned to “get the work done.” The programming capability is limited to
99 program steps and two logic compares (X≤Y? & X=0?). Many of the nice programming features of
more capable calculators are missing: absolute value, flags, increment or decrement for looping, indirect
addressing, labels, sign, subroutines, etc. This feature is further illustrated with the program examples in
appendix C.
The HP-12C is accurate: Most people who use an HP calculator expect that the answer is correct. The
situation is that this is not the case in the real world, and HP calculators are renowned for their accuracy.
There are two issues of accuracy involved in the HP-12C. First is the normal issue of making calculations
accurate rounded to ± 1 count in the 10th place. The second accuracy issue is that of the applications
programs, especially the TVM. Solutions are iteratively solved and the process may be error prone. The
algorithms used are critical and the HP-12C uses the best. The critical feature (and characteristic) of the
HP-12C is that a problem solved on the first (1981) calculator will get the same answer on the most recent
(2011) calculator. This is quite remarkable in the calculator world – and another first.
What has happened in 30 years?

While the HP-12C looks much the same on the outside it has changed on the inside over the years. One
physical specification was previously mentioned, weight. We weighed HP-12C serial number
2620A09672(8) at 117.3 grams (4.1 oz.) with 3 button cells. The 4 oz. weight has been consistent over the
32 years of manufacturer.
Major changes have been made to the electronics circuits.

The very first HP-12Cs were wrapped in a black plastic
ESD(9) film as shown in Fig. 4.
The original microprocessor is no longer made and a newer

microprocessor had to be used. Newer technology is
considerably faster and essentially the new microprocessor
emulated that of the original so the original ROM code
could be used. This is one technology reason the math has Photo by Geoff Quickfall
been consistent throughout its life. Another is reason is Fig. 4 – Early HP-12C circuit board shown wrapped
HP’s policy of maintaining a standard. in a black plastic ESD film. .
While the insides have changed and the outside looks much the same there are subtle mechanical (case)
changes that have been made over the years. HHC 2011 attendees will be asked to bring their HP-12Cs to
the conference so that we may photograph them.

The original HP-12Cs utilized gold plated circuit board traces. This practice was quickly discontinued in
favor of more current design practices. An urban legend says that the display bezel was also gold plated.
This is not true. Trivia question. There has been an HP calculator model that was gold plated on the
outside. What is it?
The HP-12C has been in continuous production for 32 years. Research was made and while this is far
from being 100% accurate it illustrates how things have changed over three decades. The HP-12C
production was in the countries listed in Table 1 below. This data was obtained from interviews with
retired HP personnel, serial numbers from friends and the Internet, and Don O’Rourke of
http://internationalcalculator.com/ The serial numbers of HP-12Cs made in China have changed in the
last few years and serial number changes and inconsistencies have been observed by the HP User
Community. One conclusion that may be made is that HP had a significant overlap of countries whenever
a country change was made to insure a continuous supply.
Table 1 – Country of HP-12C Manufacturer with Estimated Dates
1. USA1 SN ’A” September* 1981 to 2003

2. Singapore SN ”S” September* 1981 to 1996?
3. Brazil2 SN “B” 1984 to early 1992.
4. Malaysia (Indonesia)3 SN “MY” Early 1990 to late 1999
5. China SN “CNA” 1999 to Present (9/2011)
* - Obviously calculators were made several weeks or months prior

(8)
to this introduction date. Earliest known (date) 12C SN 2031A#####
= Made in USA July 4,1980
Notes:
1. The HP-12C was manufactured in the US and Singapore when introduced.
2. The HP-12C was manufactured in Brazil and because government regulations
required a Brazilian component to the machine if it was to be sold there it was
moved. Otherwise the source code would have to be provided to the Government.
3. Plant closed due to political instability.
In addition to the technology the competition has changed greatly over the last 30 years and this has
influenced the price(10) of the HP-12C. The current HP website price is $70, less than half of the original
introductory price of $150.
Even business people need to play.

The programmability of the HP-12C provides the capability to key in games programs in addition to
complex financial procedures. Numerous games have been programmed. The programs Sum of the
digits Game, BlackJack, Slot Machine, and Eleven-Thirty Game, may be found in Appendix C.
A penny for your thoughts

The exceptional accuracy of the HP-12C may be illustrated with a simple problem suggested by William
Kahan (11). “A bank retains a legal consultant whose thoughts are so valuable that she is paid for them at
the rate of a penny per second, day and night. Lest the sound of pennies dropping distract her, they are

deposited into her account to accrete (accrue) with interest at the rate of 10% per annum compounded
every second. How much will have accumulated after a year (365 days)?” Solve this on your financial
calculator and see what answer you get.
Enter the data:
n : = 60 * 60 * 24 *365 = 31,536,000 seconds per year.
i : = 10/n = 0.000 000 317 097 9198 % per second.
PV : = 0
PMT : = -0.01 = one cent per second to the bank.
FV : ?
Pressing [FV] should display one year’s accretion, but different financial calculators display different
amounts. Here is the result if you use an HP-27, HP-92, HP-37, HP-38, and HP-12C = $ 331,667.0067.
As an aside, the new HP 10bII calculates $ 331,667.006691.
Less accurate, non-HP, calculators may get answers like: $ 293,539.16035, $ 334,858.18373, or $
331,559.383549.
Observations and conclusions

The HP-12C financial calculator has been in continuous production for over 30 years –
unchanged! The hardware has changed with improved engineering practices and microprocessor
changes, but the exceptional accuracy has been maintained and a problem solved on a 1981
machine will get the same answer as one solved on a 2011 machine. This is reasonable because
there is only one correct answer. We have tried to give the reader an overview of this exceptional
calculator in this short review and we both look forward to its 35th anniversary.
Additional HP-12C information and support may be found in the appendices.
Appendix A - 1982 HP Catalog Page Showing New HP-12C
Appendix B - 1981 HP Sales Tip Brochure
Appendix C - Four HP-12C Game Programs
Appendix D - HP-12C References
About the Author
Gene Wright is the author of textbook "Quantitative Analysis for Business", a
business math textbook using the HP-10BII and HP-12C, available from
Amazon.com. He is also a video lecturer for a CFA exam review course. A
former teacher at Lipscomb University in Nashville, Tennessee, he now works
for a consumer electronics company. Gene has written many articles on HP
calculators and serves on the annual HHC committee.
About the Author
Richard J. Nelson has written hundreds of articles on the subject of HP’s
calculators. His first article was in the first issue of HP 65 Notes in June 1974.
He became an RPN enthusiast with his first HP Calculator, the HP-35A he
received in the mail from HP on July 31, 1972. He remembered the HP-35A in
a recent article that included previously unpublished information on this calcul-
ator. See http://hhuc.us/2007/Remembering%20The%20HP35A.pdf
He has also had an article published on HP’s website on HP Calculator Firsts.
See http://h20331.www2.hp.com/Hpsub/cache/392617-0-0-225-121.html.

Notes: The HP-12C, 30 Years & Counting
(1). A list of other HP Calculator firsts may be found at: http://h20331.www2.hp.com/Hpsub/cache/392617-0-0-225-121.html
(2). The countries in historical order are USA, Brazil, Singapore, Malaysia, and China.
(3). The initial specifications of the 12C may be found on page 646 of the 1982 hardbound catalog reproduced in
Appendix A. The price was $150. Copies of these catalogs may be found at: http://www.hpmuseum.net/exhibit.php?content=HP%20Catalogues#
(4). A conference reviewer, Jo Vandale of Belgium, wrote in a 9/23/2002 Conference review regarding the HP-
12C session, "The screen will be readable and the form factor will be horizontal ... at least for the HP12C
successor".
(5). The HP-12C has had many engineering changes to reduce cost and to utilize current electronics technology.
These changes have not substantially changed the weight which has been consistently specified by HP as 4 oz.
(6). Battery life is somewhat related to the quality of the
manufacturer of the battery. When the button cells eventually
need replacement silver button cells are frequently used. See
the figure at the right for a comparison.
(7). The UK HPCC publication Datafile has an article and
program to add Sin, Cos, Tan and their inverses (in radians) to
an HP-12C. The title is HP-12C Tried & Tricky Trigonometrics
by Valentin Albillo. It is in the January/February 2002 Issue -
V21N1p12.
(8). The coded serial number (SN 2620A09672) is decoded as follows. The first two digits are the years since 1960
= 1986, the next two digits is the week of the year = mid April, “A” is the country of manufacturer (USA), and the
remaining five digits are sequential for the week = at least 9,672 for that week. The date is not the date of actual
manufacturer, but it is led by some number of weeks (4 – 6) for the calculator to be shipped and sold. The reason
for this is the warranty period is based (back then) on the serial number – or a copy of your sales receipt. For
additional details on decoding HP Calculator serial numbers see: http://www.hpmuseum.org/collect.htm#numbers
(9). ESD is ElectroStatic Discharge and is most commonly experienced when you walk across a carpet (especially
one made of synthetic materials) and you get sparked by touching a large metal object. Modern electronics circuits
are very sensitive to ESD because a single “spark” will destroy them in microseconds.
(10). You may find the Current HP-12C Data Sheet and pricing at:
http://h10010.www1.hp.com/wwpc/us/en/sm/WF06a/215348-215348-64232-20036-215349-33525.html
Very few calculators have more than a ten year history
of being manufactured unchanged. A three decade span
is truly exceptional, and the list price of the 12C
illustrates that prices may even increase under unusual
market conditions. The first 20 years were essentially
pre-Internet. Once the Internet became the primary
source of information the discount pricing history
became more obscure. The data that was found is shown
at the right. Early sources were the HP Catalog and all
75 EduCALC catalogs. Both are no longer published.
Even with the price increases in 1989 and 1996 the HP-
12C popularity did not change. HP suggested list price history for the HP-12C.
(11) William Kahan of the University of California Berkeley consulted with HP on the HP-12C and other
calculators (HP-15C & HP-34C) to insure algorithmic accuracy. He has a long and productive career in

promoting computational accuracy in computers and calculators. See page 15 of Mathematics Written in Sand,
Version of 22 November, 1983 at: http://www.cs.berkeley.edu/~wkahan/MathSand.pdf.
Appendix A –1982 HP Catalog Page Showing New HP-12C

Appendix B –1981 HP Sales Tip Brochure - Page 1 of 2

Appendix B –1981 HP Sales Tip Brochure - Page 1 of 2

Appendix C – Four HP-12C Game Programs - Page 1 of 5

Appendix C – Four HP-12C Game Programs Cont’d - Page 2 of 5




Appendix D – HP-12C References - Page 1 of 1
Here is a small sampling of HP and third party support books, manuals, and Internet links.
D1 - http://h10032.www1.hp.com/ctg/Manual/c00363319.pdf
D2 - http://h10032.www1.hp.com/ctg/Manual/c00367122.pdf
D3 - http://dl.dropbox.com/u/10022608/HP-12CRealEstateApplicationsHandbook.pdf
HP-12C Leasing Applications handbook - http://dl.dropbox.com/u/10022608/HP-12CLeasingApplicationsHandbook.pdf
HP has learning module “training guides” that address a wide range of topics. Here is one that explains
how to write a small 12C program. - http://h20331.www2.hp.com/Hpsub/downloads/HP12Cprogram.pdf
Here is zip file of all of them. http://h20331.www2.hp.com/Hpsub/downloads/12c.zip
Fig. D1 – Handy sized Owner’s Fig. D2 – Solutions handbook Rev. F Fig. D3 – Real Estate Applications
Handbook Rev. G 11/85. 7/87. Handbook Rev. B 3/84.
Early third party support booklets are represented below. Chary is still active in HP-12C support using
videos. D3 & D4 - http://hp-12c-calculator.com/?page_id=60
Gene’s business math book is at: http://www.amazon.com/Quantitative-Analysis-Business-Statistics-Calculator/dp/1888840382/
Fig. D3 – Chary Software Program- Fig. D4 – Chary Software Program- Fig. D6 - Renaissance Publications.
ming Hints book Rev. A 12/85 ming Hints book Rev. A 8/86 Problem examples book © 1985

Calculator Programmability—How Important is it in 2011?
Return to Top
U UH HU U
Calculator Programmability – How Important is it in 2011?
Richard J. Nelson
Introduction
When the non-programmable scientific HP-35A calculator first appeared many HP
enthusiasts wrote down keystroke lists(1) on index cards for the most efficient method of
solving problems, especially those that required iterating to a solution. Two years later
the programmable HP-65A appeared and tens of thousands of HP calculator users
became programmers. This was in the mid 70’s and the HP-65A was the first personal
computer that gave programmability to everyone, not just to college students or
employees of large corporations that could afford room sized computers.
The scientific programming language of the institutional computer was a higher-level

Fig. 1- HP-65A. language, typically FORTRAN or BASIC. The language of the HP Calculator was a
lower level keystroke programming language that was “officially” named FOCAL(2) with the peak of the
HP-41 popularity of the late 70’s and early 80’s. The HP-28C, in mid 1980, offered additional
programmability in terms of its higher-level RPL programming “language.” The vast majority of legacy
RPN users didn’t like the new language and the number of users who programmed in RPL vs. FOCAL
dropped by a factor of 100. The new HP 15c Limited Edition, LE, offers a FOCAL like programming
language and it will be used as an example.
Programming, what is it?

A program is a series of instructions that are usually stored in memory to be “executed”
multiple times. The term programming is also applied to the operation of modern
electronic devices in terms of setting the various options of how the device works.
Programming, as used here, is a list of keystrokes and operations that the calculator
Fig. 2 - Binary. executes in the order as programmed into memory. A simple short program that solves
a specific task that is used on a daily basis in business or on the production line can save many hours per
week in computational time.
A program may also “manage” a data base and serve as a ‘look up” mechanism for information such as
prices, specifications, or any numeric data in the form of tables or equations.
Programming Advantages
What are the advantages of programmability for the calculator user?
1. Programming provides a quick simple method of solving the same problem over and
over again. A single keystroke could replace dozens of keystrokes. An example is
calculating the volume of a dozen grapefruit from a quality sampling.
Fig. 3 - Languages 2. A program could solve more complex problems that the user couldn’t solve
otherwise. The program is written by an expert and used by anyone. An example is a complex
statistical calculation such as mean and standard deviation.
3. A program could contain data that isn’t conveniently looked up or remembered. Making English
Metric conversions is a simple example of this program usage e.g. converting grains to grams or troy
ounces to grams.

4. A program could address a very specialized problem that information would be difficult to research.
An example is a program that calculates aircraft altitude based on pressure that uses a very obscure
equation.
5. A program could implement very complex mathematical methods that require thousands of loops or
iterations to arrive at a solution. An example is a random walk calculation for a complex process.
6. A program could apply “trial and error” and data search methods to arrive at a solution. Finding a
root of an equation or multiple equations is an example.
7. A program could act as a simple database where pricing information from part number inputs is either
looked up or calculated based on quantity or delivery time inputs. An example is part number
conversion to a data base number that has the desired price.
8. A simple program could utilize several applications programs that are part of the calculator to arrive
at an optimum solution saving keystrokes similar to #1 above. An example is using a symbolic
solver, unit management, unit conversion, and a plotting application to see a complex relationship.
9. A program could automatically label and identify inputs and outputs for user convenience. This is
especially useful for multiple problems solutions on the stack.
10. Programmability provides a creative applications tool for the user community to develop and share
knowledge and techniques for more effective problem solving. High precision arithmetic programs
provide an infrequently used tool for those situations that require a high number of significant digits.
11. A creative program could be adaptive and self-learning to the extent that it exceeds the ability of its
user. Game playing strategies have been programmed for the HP-65A and later machines so that
after many games the machine is able to beat its human competitor.
12. Having a programmable machine provides a research tool for the manufacturer for the next
generation of calculators. Many of the built-in applications of more advanced machines were
applications programs created by HP’s customers. A good example is the HP49 and HP50 models
with added applications that were originally programmed by HP’s users for the HP48.
Programming online or offline?

High end scientific or graphing calculators may have an input/output, I/O, capability to
connect to a computer - usually it is USB. When this capability is available software is
often developed to write calculator programs on the computer. Most users will write
the program on the calculator, but more elaborate and complex programming is more
effectively done on the computer because of the time it saves. Usually this dividing
Fig. 4 – Connect. line is between the amateur who writes an occasional program on the calculator and the
professional who develops commercial calculator software on the computer.
HP 15c LE programming
The HP 15c RPN program instruction set has a respectable collection
of programming features. The large number of applications
programs, especially those that may be adapted from many other
Fig. 5 – HP-15c+ Program Mode. RPN programmable calculators is a substantial advantage
the exceptional speed of the HP 15c LE provides an additional advantage of this programmability.

An important marketing aspect of the HP 15c LE is promoting the advantages of its being programmable
and providing resources for obtaining programs. See the document titled HP-15C LE Speed Comparison
and Reference provided at HHC MMX (2010) as a resource(3). The HP 15c LE is on the order of 125 to
175 times faster than the original 1982 HP-15C. The “Comparison and Reference” document provides a
table of speed comparisons, a list of references for programs, and a Quick Reference Guide for the
features and functions of the HP-15C. Also see Appendix A for links to additional resources.
Off line program storage

The greater the number of features, operations, or functions of a
calculator the greater is the need for it being programmable.
Programming increases the power and capability of the calculator and
Fig. 6 – Off line storage is needed. memory capacity then becomes an issue. How many programs need to be
stored and available to the user? When the number of programs increases, another aspect also becomes a
vital issue. How may the programs be protected? The HP 15C has continuous memory as indicated by
the “C/c” in the model number. There is no off line storage for HP-15C or HP 15c LE programs. This
may be viewed by modern calculator users as a serious limitation. The larger the program memory the
greater the need for offline storage, e.g. a USB port.
Calculator Programmability Observations and Conclusions

Modern machines of the last ten years take advantage of lower memory cost (for
functions and programs) to keep the cost of the calculator low. The primary advantage
of the dedicated calculator is low cost and maximum convenience.
More complex problems are often solved on a personal computer, which tends to be
turned on for extended periods, such as a work week of five days. Comparing the
Fig. 7 – Tools. calculator solution with a computer solution doesn’t yield a clear advantage in turn-on
convenience (instant on vs. boot time). Still, a single key press for a programmed solution is nearly
impossible to improve and the usefulness of programmability is easily justified in the midrange to the
high end models for most of the dozen reasons listed above.
The programming method, however, must be as simple as possible so the average user may easily master
it. The HP 15c LE with its FOCAL like keystroke programming is easy to learn. Even if the student is
less interested in programming their own machine he or she will still need those programs suitable for the
course at hand. Being able to share programs via the Internet and down load them via a USB port will
always be a justification to include the programming feature of any serious scientific or graphing
calculator.
The large number of features, memory size, and super speed of the HP
15c makes its programmability an important feature when compared to
other calculators - for the 12 programming reasons listed above.
To answer the title question, “Calculator Programmability – How Important is it Today?” I would have to
say that it is as important as it ever was. A more modern consideration is having an I/O capable of down
loading the programs regardless of the language of the operating system and even if most users are not
writing programs themselves.

___________________________________________________________________________________________
Notes - Calculator Programmability – How Important is it Today?
1) The most famous collection of these “index card programs” was written by Lee Skinner of Albuquerque New
Mexico who sold them to HP. Readers may know these as the HP-35 Math Pac or the similar HP-45 Math
Pac. See page 5 of: http://hhus.us/2007/Remembering%20the%20HP35A.pdf for additional details.
2) FOCAL is believed to be what HP calls the Forty-One CAlculator Language. See HP Solve issue 23 for a
discussion of FOCAL. Is this an HP Urban legend?
3) You may obtain copies of the (must have) Reference DVD at: http://www.pahhc.org/ppccdrom.htm
Appendix A – HP-15C Programs in the Owner’s &

Advanced Functions Handbooks
Fig. A1 – Owner’s Handbook (initial1982 version) and Advanced Functions handbook (later 1984 version).
Owner’s Handbook - 3/82. (00015-90001) http://hp15c.org/hp15c.pdf

1. Example Section 6: Calculating base area, volume, and surface area of cans, page 72,
2. Example Section 6: Calculating 5x4 +2x3, page 80.
3. Example Section 8: Figure the radio-activity at 3-day intervals until a given limit, page 93.
4. Example Section 8: Calculate payments of an investment for college tuition, page 95.
5. Example Section 8: Converting temperature inputs in degrees Celsius or Kelvin, page 99.
6. Example Section 9: Calculate the slope of the secant line of a parabola, page 103.
7. Example Section 10: Modifies Example #3 to store intervals into registers using DSE, page 113.
8. Example Section 10: Demonstrate FIX displays for all decimal positions, page 114.

Appendix A – HP-15C Programs in the Owner’s &
Advanced Functions Handbooks – continued.
9. Appendix D: Calculate the shear stress, Q = 3x2 – 45x2 + 350, page 228.
10. Appendix D: Investigate , Page 231.
11. Appendix D: Use SOLVE initial estimates for #10 above of 10, 1, & 10-20, page 232.
12. Appendix D: Use deflation for all roots of: 60x4 – 944x3 + 3003x2 + 617x -2890 = 0, page 234-237.
13. Appendix E: Calculate Page 242.
(1) These are primarily short keystroke routines.
Advanced Functions Handbook - REV B, 8/84. (00015-90011) http://hp15c.org/hp15cAdvanced.pdf

Only applications programs are included in this list, excluded are the short keystroke expression
evaluation routines.
1. Section 1: Time Value of Money, 108 lines, page 29.
2. Section 1: Discounted Cash flow, 068 lines, page 40.
3. Section 2: Four integrals of four special functions, 083 lines, page 61.
4. Section 3: Calculating the gamma function, 028 lines, page 66.
5. Section 3: Storing and Recalling Complex numbers in a matrix, 025, page 76.
6. Section 3: Calculating nth roots of a complex number, 037 lines, page 78.
7. Section 3: Solve and Equation for its Complex roots, 059 lines, page 82.
8. Section 3: Evaluate Complex integral, 045, page 85.
9. Section 3: Complex potential function. 061, page 90.
10. Section 4: Constructing an identity Matrix, page 119.
11. Section 4: Solves system of equations AX = B for X, then makes residual correction, page 120.
12. Section 4: Solves a system of Nonlinear Equations, pages 123 – 126.
13. Section 4: Solves a large System of Complex Equations, pages 129 – 131.
14. Section 4: Multiple linear regression (Least Squares), pages 135 & 136.
15. Section 4: Weighted least-squares multiple linear regression, pages 143 – 146.
16. Section 4: Eigenvalues of a Symmetric real Matrix, pages 150 – 153.
17. Section 4: Eigenvectors of a Symmetric Real Matrix, pages 155 – 157.
18. Section 4: Optimization (Min & Max) of a real-valued function, pages 163 – 166.
Special note: The Appendix of the Advanced Functions Handbook, pages 172 – 211 provides an excellent
discussion of the accuracy of all HP calculator numerical calculations – prior to the RPL generation. See example
in Fig. A2. There are four error evaluation and error correcting programs.
1. Calculating a more accurate λ = ln(1 + x), pages 181

& 182.
2. Calculating angle (cosine law) in a triangle, error

analysis (very nice), pages 197 – 199.
3. Accurate roots to real quadratic equation - real and

complex roots, pages 205 & 206.
4. Most accurate variation of #3, uses ∑+ & ∑-. Certain

calculations are done to 13 places, pages 208 - 210. Fig A2 – Error analysis text example page 179.

Converting Decimal Numbers to Fractions

U UH HU U
Return to Top
Converting Decimal Numbers to Fractions
Joseph K. Horn
What fraction is π equal to? Most people would reply “22/7” and of course they would be wrong, since
π is an irrational number which means that it cannot be expressed as a ratio of integers. But 22/7 is
close to π . Is there a fraction that is closer?
The answer originally flowed from the following line of brute-force observation: “ π ≅ 3 . Well, not quite
1 1
3, but more precisely π ≅ 3 . (Hence 22/7). Well, not quite 7, but 7 , which means that
7 15
1
π ≅ 3+ . (Hence the better approximation 333/106).” And so on. Following this line of reasoning,
1
7+
15
closer and closer approximations to π can be generated. The trick is to pick the correct adjustments.
There is a method for doing this that avoids brute-force searching. It is called “The Continued-Fraction
Algorithm.” Here it is.
First you make a list of numbers like this:

A) Write down the integer part (INT) of the number.
B) Subtract the integer part by pressing FRAC.
C) Reciprocate this by pressing 1/x.
D) Go to step A. Repeat a few times. Stop if you hit a big number.
This will generate the following list for π : { 3, 7, 15, 1, 293 }. Recognize the first three numbers? They
are the denominators printed in red in the large equation for π above. Try using the first four numbers
from this list instead of just the first three; it yields this larger expression:
1 355
π ≅ 3+ =
1 113
7+
1
15 +
1
Hence a better approximation for π is 355/113. Now use all five numbers from the list:
1
π ≅ 3+
1
7+
1
15 +
1
1+
293
This huge expression of nested reciprocals (called a “continued fraction”) simplifies to 104348/33215,
which is therefore a fraction that is even closer to π than 355/113. It is so close, in fact, that it evaluates
on the HP-15 to the same value as the π key.
This procedure can be used to quickly generate a fraction that approximates any decimal number to any
desired number of digits. Let’s do it by hand one more time to appreciate the method.

What fractions best approximate π ? First we follow the 4 steps A-D above and get this list: { 1, 1, 3, 2,
1, 1, 6, 1, 28 }. These numbers are called the “partial quotients” of the continued fraction. We then build
some continued fractions from the first few partial quotients, thereby deriving better-and-better approxi-
mations to π , including these three:
1 16 1 39 1 257
1+ =1 + = 1+ =
1 9 1 22 1 145
1+ 1+ 1+
1 1 1
3+ 3+ 3+
2 1 1
2+ 2+
1 1
1+ 1+
1 1
1+
6
The fractions generated by this method are called the “convergents” to the number being approximated.
The infinite set of convergents for π begins thus:
1 2 7 16 23 39 257 296 8545

, , , , , , , , , ...
1 1 4 9 13 22 145 167 4821
Since the continued fraction algorithm is mathematically well-defined, it can be programmed on any
programmable HP calculator. Appendix A of this article is an HP-15C program that implements it.
Appendix B is an HP 35s program. The HP-15C program only needs 45 steps to convert any decimal
number into a fraction with any desired accuracy. Just input the decimal number, set the display FIX to
control the accuracy, and GSB A. The numerator appears. Press X<>Y to see the denominator.
Optionally press ÷ RCL – 0 to see the difference between the input decimal and the output fraction.
5 +1
The input that makes the program work the hardest is the golden ratio, , because its partial
2
quotients are { 1, 1, 1, 1, 1, 1, 1, … }. The HP-15C program says that the fraction that best approximates
6765
the golden ratio to 7 digits is . What are its other convergents? It has more than any other number!
4181
If you write the first several down, you will see the Fibonacci Series hidden in it. Many other Number
Theory goodies await discovery along the path of the Continued Fraction Algorithm. Happy exploring!
About the Author

Joseph K. Horn is a high school math teacher in Orange County, California. He
has authored many articles related to HP calculators in addition to his book titled,
HP-71 Basic Made Easy. He is sometimes known in the HP user community as
“The math bug hunter.”
Joseph serves on the HP Handheld conference, HHC, committee and is webmaster

for the HHC website. hhuc.us

Appendix A - Decimal to Fraction for the HP-15C
Joseph K. Horn
Version 1.2
Converts any positive decimal number to a fraction.
Accuracy is controlled by the current FIX setting.
01*LBL A 42 21 11 16 X<>Y 34 31 1/x 15
02 STO 0 44 0 17 ÷ 10 32 STO 1 44 1
03 STO 1 44 1 18 RND 43 34 33 GTO 0 22 0
04 0 0 19 RCL 0 45 0 34*LBL 1 42 21 1
05 ENTER 36 20 RND 43 34 35 RCL 2 45 2
06 ENTER 36 21 TEST 5 43 30 5 36*LBL 2 42 21 2
07 1 1 22 GTO 1 22 1 37 ENTER 36
08 R↑ 43 33 23 + 40 38 ENTER 36
09*LBL 0 42 21 0 24 CLx 43 35 39 RCL 0 45 0
10 INT 43 44 25 RCL 2 45 2 40 × 20
11 R↑ 43 33 26 ENTER 36 41 . 48
12 × 20 27 ENTER 36 42 5 5
13 + 40 28 R↑ 43 33 43 + 40
14 STO 2 44 2 29 RCL 1 45 1 44 INT 43 44
15 GSB 2 32 2 30 FRAC 42 44 45 RTN 43 32
Instructions:
(1) Use a FIX setting as desired.
(2) Place decimal number in X.
(3) GSB A ➯ see numerator.
(4) X<>Y ➯ see denominator.
Optional: ÷ RCL - 0 ➯ “error” (difference between input and output)
Example:
Convert π to a fraction accurate to 5 decimal places.
(1) FIX 5
(2) π
(3) GSB A ➯ 355
(4) X<>Y ➯ 113
Answer: 355/113
Notes:
(a) The program uses registers 0 through 2, so make sure that enough register memory is allocated.
(b) As always, turning off USER mode makes it easier to key in the program.
(c) When the program finishes, register 0 contains your original input. Registers 1 and 2 are used for
temporary storage.
(d) The Golden ratio, 1.618033989, is the most challenging (takes longest time).

Appendix B - Decimal to Fraction for the HP-35s
Joseph K. Horn
Note: Program execution is the same as for the HP-15C program except it is launched by XEQ A ENTER
instead of GSB A.
No attempt was made to optimize the program by using line number addressing instead of labels.
# Prgm. Instr. # Prgm. Instr. # Prgm. Instr. # Prgm. Instr.

1 A001 LBL A 12 B004 × 23 B015 + 34 C001 LBL C
2 A002 STO A 13 B005 + 24 B016 CLx 35 C002 RCL C
3 A003 STO B 14 B006 STO C 25 B017 RCL C 36 D001 LBL D
4 A004 0 15 B007 XEQ D001 26 B018 ENTER 37 D002 ENTER
5 A005 ENTER 16 B008 x<>y 27 B019 ENTER 38 D003 ENTER
6 A006 ENTER 17 B009 / 28 B020 R↑ 39 D004 RCL A
7 A007 1 18 B010 RND 29 B021 RCL B 40 D005 ×
8 A008 R↑ 19 B011 RCL A 30 B022 FP 41 D006 0.5
9 B001 LBL B 20 B012 RND 31 B023 1/x 42 D007 +
10 B002 INTG 21 B013 x=y? 32 B024 STO B 43 D008 INTG
11 B003 R↑ 22 B014 GTO C001 33 B025 GTO B001 44 D009 RTN
Note: # column numbers do not appear in the display and are included for reference only.

Limited Edition HP 15c Execution Times

U UH HU U
Return to Top
Limited Edition HP 15c Execution Times
Namir Shammas
Introduction
The famous French writer Voltaire seems to have captured the driving force behind progress when he said
“The better is the enemy of the good.” Recently, Hewlett-Packard launched the HP 15c Limited Edition
(which I will nickname the HP-15C+ in this article) due to popular demand by users of vintage HP
calculators. The vintage HP-15C was the second calculator to support the Solve and Integrate features,
after the HP-34C, and the first one to support both complex math and matrix operations. The HP-15C has
earned many fans that went online and posted a petition for HP to re-launch that machine. After many
years of waiting, I am happy to report that HP has listened and indeed responded to these very loyal HP-
15C fans. The HP 15c Limited Edition is here and comes with the same features of the original HP-15C,
with added speed. I mean this little puppy is fast! How fast, you may ask? This article answers this
question by comparing the speeds of the old HP-15C and the new HP 15c.
Speed Test Setup

Like many HP enthusiasts, I heard rumors about the speed of the new HP-15C+. I decided to investigate
the speed of the new machine and compare it with the vintage version. My testing includes the following
aspects:
1. Each test uses a program that repeats a set of operations for a fixed number of times. Timing
(using the stopwatch of an HP-41CX) starts and stops when the test program starts and stops,
respectively. To help the setup a test, I use code that initializes memory registers used in the test
program.
2. I use simple programs that execute empty loops to obtain the time for these loops. When I perform
a test for an operation set, I subtract the empty loop time to obtain the time for the operation set. I
calculate the time needed to perform that operation set by dividing the number of test loops by the
net operations’ time. I get the speed of an operation set.
3. To calculate the speed ratio, I divide the speed of the operation sets performed on the HP-15C+ by
that of the HP-15C. This is the target statistic that I am looking for in each test.
4. To test functions that need arguments, I initiaize the tests by storing the arguments in memory
registers. The test loops then recall the values from the registers and supply them as arguments to
the tested functions. This approach offers uniform timing to obtain the arguments of a function.
While most functions need one argument, some like Y^X need two. I have also timed loops that
recall one memory register and also two memory registers. I subtract the time for these loops from
the time of the loops that recall arguments for built-in functions and then execute these functions.
5. Since the HP-15C and HP-15C+ vary in speed significantly, I use loops that differ in the number
of iterations to keep testing within reasonable times and minimize human errors. I estimate human
error to be between 0.1 and 0.2 seconds.
6. The DC voltages for the cells in the vintage HP-15C were read by an AVO meter to be 1.505,
1.509, and 1.503 volts. The DC voltages for the two cells of the HP-15C+ were measured as 2.97
and 2.98 volts. The tests were run at room temperature of 70 degrees Fahrenheit.
Speed Test Code

Here is the code for an empty loop that iterates 100 times on the HP-15C:

LBL A
EEX
2
STO 0
RTN
LBL B
DSE 0
GTO B
RTN
Label A initializes the loop counter stored in register 0. The initial loop counter value depends on the test
and the machine version. Label B performs the test to time an empty loop. I use the DSE command
because I can count down from numbers greater than 1000. If I use the ISG command instead, I am
limited to a maximum of 1000 iterations. On the HP15C+ I use tests programs that loop10000 times.
This high loop iteration justifies selecting the DSE command to control the loop iterations.
Here is the test that recalls a value from register 9:

LBL B
RCL 9
DSE 0
GTO B
RTN
Keep in mind that the above loop recalls a value from register 9 AND also pushes the stack. Most the
tests involve operations and/or calculations that push the stack and/or store a value in the LastX register.
A few operations, like STO n, do not move the stack and do not alter the Last X register. Math operators,
like +, drop the stack and also update the Last X register. Single-argument functions alter the X register
in the stack and update the LastX register. The RCL n operations push the stack up but do not affect the
LastX register.
Here is a sample code that calculates the sine of a 0.6 radians, stored in register 9:
LBL A
RAD # switch to radians
.
6
STO 9 # store 0.6 in memory register 9
EEX
2
STO 0 # initialize the loop counter for the HP-15C
RTN
LBL B
RCL 9 # recall argument value
SIN # calculate the sine
DSE 0
GTO B # end of the test loop
RTN

Label A initializes the test by switching to radians, storing 0.6 in memory register 9, and setting up the
loop counter. Label B performs the test by repeatedly recalling the value of 0.6 from memory register 9
and executing the sine function. The above code is very similar to versions for the other built-in functions.
In the case of a using function with two arguments I use code that looks like the following one:
LBL A
Π
STO 9 # store value for first argument
STO 8 # store value for second argument (same as first)
EEX
2
STO 0 # initialize loop test counter for the HP-15C
RTN
LBL B
RCL 9 # recall first argument
RCL 8 # recall second argument
Y^X # use built-in function
DSE 0
GTO B
RTN
Label A initializes the arguments by storing them in registers 8 and 9, and then initializes the test loop
counter. Label B performs the test by repeatedly recalling the two arguments from registers 9 and 8, and
then using the built-in function Y^X.
Testing the Σ+ operation is a bit different. The test repeats the process of adding 100 random values for Y
and X in the statistical registers. The test uses the following code:
LBL A
EEX
2
STO
0 # initialize the loop counter for the HP-15C
RTN
LBL B
CLEAR Σ # clear the statistical summations
EEX
2
STO 1 # set counter for the inner loop
LBL 0 # start inner loop
RAN # # generate random Y
RAN # # generate random X
Σ+ # add to statistical summations
DSE 1 # test end of inner loop
GTO 1 # end of inner loop
DSE 0 # test end of outer loop
GTO B # end of outer loop
RTN

Label A initializes the main loop in label B. This label has an inner loop that starts at label 0. The first
lines after label B initialize the statistical summations and the counter for the inner loop. The inner loop
generates the random values for Y and X, and then adds them to the statistical summations. The outer
loop repeats the process of generating random numbers and adding them to the statistical registers.
Special Speed Test Code

Most of the tests I conducted use programs that are similar to the ones I presented in the last section.
These programs are aimed at simple operations and the evaluation of built-in functions. In this section, I
present the code for more complex tests. These tests are:
• Calculating e^π using a Taylor expansion polynomial. The computations add the values of
different Taylor expansion terms to obtain the exponential value. When the program rounds the
value of a term to zero (based on the current number of displayed digits), the iteration stops. I
tested calculating e^ π in FIX 7 and FIX 4 display modes.
• Calculating the integral of 1/x from 1 to 100.
• Calculating the root of the function e^x-3*x^2 between 3 and 4.
• Converting decimals to fractions. This test is based on a program written by Joseph Horn . The
test uses the value of the golden ratio, since it requires more calculations than many other common
numbers with fractions, like π or e. The test program uses the FIX 9 display mode.
The code in this section is aimed at the new HP-15C+. The code for the vintage HP-15C differs only in
the values stored in the loop counter (in register 0), which is set to 10.
The code for calculating e^π using Taylor approximation is:
LBL A
FIX 7 # can set to other values
Π # π
STO 1
EEX
3
STO 0 # initialize the loop counter for the HP-15C+
RTN
LBL B
0
STO 2 # counter
STO 3 # sum
1
STO 4 # x^counter
LBL 1 # start of inner loop
RCL 4
RCL 2
X!
/
STO+ 3
RND # round term

x=0? # term rounds to zero?
GTO 2
1
STO+ 2 # counter += 1
RCL 1
STO* 4 # x^counter
GTO 1 # end of inner loop
LBL 2
RCL 3 # recall e^π
DSE 0
GTO B
RTN
Here is the code for testing the Integrate function:
LBL A
1
STO 9 # store first argument
EEX
2
STO 8 # store second argument
EEX
3
RTN
LBL B
RCL 9 # recall lower limit for integration
RCL 8 # recall upper limit for integration
ʃ 0 # integrate function in label 0
DSE 0
GTO B
RTN
LBL 0 # custom function to integrate
1/X
RTN
Here is the code for testing the Solve function:
LBL A
3
STO 9
4
STO 8
EEX
3
RTN
LBL B

RCL 9 # recall first value to define solution interval
RCL 8 # recall second value to define solution interval
SOLVE 0 # solve function in label 0
DSE 0
GTO B # end of loop
RTN
LBL 0 # custom function to solve
e^X
LSTX
X^2
3
*
-
RTN
Here is the code for the decimal to fraction program (version 1.1) that Joseph Horn wrote:
LBL A
STO 0
STO 1
0
ENTER
ENTER
1
R^
LBL 0
INT
R^
*
+
STO 2
GSB 2
X<>Y
/
RND
RCL 0
RND
TEST 5
GTO 1
CLx
+
CLx
RCL 2
ENTER
ENTER
R^
RCL 1
FRAC
1/X

STO 1
GTO 0
LBL 1
RCL 2
LBL 2
ENTER
ENTER
RCL 0
*
.
5
+
INT
RTN
I am including the above version, because that’s the one I used for testing, and in case Joseph Horn
further improves his code. Here is the code I added to Joe’s program to test the decimal to fraction
calculations:
LBL C
FIX 9
STO 4 # store number (e.g. golden ratio)
EEX
3
RTN
LBL B
RCL 4
GSB A # call Joe Horn’s program in LBL A
DSE 3
GTO B
RTN
LBL C initializes the test. Enter the value for the number with a fractional part and then press the key C.
In the case of the golden ratio, enter 1.6180339887 and then press the key C. Label B starts the test by
repeatedly recalling the value with a fractional part from register 4 and then invoking Joe’s code in label
A as a subroutine. The test code uses register 3 as a loop counter since Joe’s code already uses registers
0, 1, and 2.
Results
Table 1 shows the speed comparison summary. The table has the following columns:
1. The Test column specifies the test conducted.
2. The HP-15C column lists the speed for the various operation sets for the HP-15C.
3. The HP-15C+ column lists the speed for the various operation sets for the HP-15C+.
4. The Speed Ratio column displays the ratio for the speeds of the HP-15C+ divided by the speeds
of the HP-15C.
Tables 2 and 3 show the detailed timings for the HP-15C and HP-15C+, respectively.

Table 1. List of timing tests.
Test HP-15C HP-15C+ Speed Ratio
Result Result
Ops/sec Ops/sec
Empty loop 100 iterations 3.06
STO 9 13.84 2475.25 178.84
RCL 9 12.49 2252.25 180.29
RCL 9, RCL 8 6.20 1119.82 180.57
RCL 9, RCL 8 for 10 iters 6.21 1176.47 189.41
Roll entire stack down using
four Rv commands 3.58 595.24 166.04
+ 10.47 1572.33 150.24
- 10.52 1666.67 158.42
* 8.58 924.21 107.72
/ 6.64 750.19 113.02
997, SQRT 6.58 734.21 111.53
997, X^2 8.61 1112.35 129.14
π, 1/X 6.94 766.87 110.43
π, π, Y^X 0.97 105.85 108.68
SIN(0.6 RAD) 1.18 123.32 104.18
COS(0.6 RAD) 1.25 133.05 106.52
TAN(0.6 RAD) 1.89 211.60 111.76
ASIN(0.6) 1.48 173.04 116.68
ACOS(0.6) 1.41 163.85 115.83
ATAN(0.6) 2.26 263.09 116.23
SINH(0.6) 1.63 182.68 111.86
COSH(0.6) 3.58 227.38 63.56
TANH(.6) 1.24 131.35 106.24
ASINH(π) 1.18 122.46 104.00
ACOSH(π) 1.25 132.82 105.92
ATANH(0.6) 2.32 284.09 122.36
45, 45, ->P 2.45 303.21 123.89
π/4, 63,->R 0.96 98.84 102.48
1.3333, ->H 9.44 1412.43 149.58
1.5591666667, ->HMS 11.57 1805.05 155.96
LN(997) 3.65 426.44 116.67
LOG(99 ) 2.74 290.61 105.99
EXP(2.123) 2.67 308.83 115.87
10^2.123 2.69 305.62 113.78
π, X! 0.51 59.21 116.53
Random number 7.59 859.85 113.28
2 Random numbers 3.76 430.29 114.44

Test HP 15C HP-15C+ Speed Ratio
Result Result
Ops/sec Ops/sec
FRAC of π 13.23 2141.33 161.88
INT of π 13.61 2227.17 163.70
Integrate 1/x from 1 to 100 3.04E-02 3.63 119.71
Solve (e^x-3*x^2) for
guesses 3 and 4 8.43E-02 10.95 129.97
Adding pairs of rand m
numbers in stat summations 1.78 205.76 115.76
Calculate e^π using FIX 7 3.75E-02 5.75 153.44
Calculate e^π using FIX 4 4.76E-02 7.37 154.83
Fractions for golden ratio in
FIX 9 1.52E-02 2.47 161.83
Looking at Table 1, you can notice the following speed ratios for the different categories of operators,
functions, and programs:
• The STO and RCL operations show the highest speed ratio gain which is around 180.
• The stack roll down operation shows a high speed ratio of about 166.
• The addition and subtraction operators show speed ratios that average 154.33.
• The multiplication and division operators show speed ratios that average 110.
• The square root, square, reciprocal, and raising to power operations have speed ratios that average
114.95.
• The trigonometric functions and their inverses have speed ratios that average 111.87.
• The hyperbolic functions and their inverses have speed ratios that average 102.32.
• The conversion to polar coordinate has a speed ratio of 122, while the conversion to rectangular
coordinates has a speed ratio of 102.
• The Hour and HMS conversion functions have speed ratios that average 152.77.
• The common log, natural log, and their inverse functions have speed ratios that average 113.08.
• Random number generation has a speed ratio of 113.28.
• The INT and FRAC functions have speed ratios that average 162.79.
• The Integral and Solve functions have speed ratios that average 124.84.
• Adding random pairs of X and Y values to statistical summations is an operation with a speed
ratio of 115.76
• The programs that calculate e^π have speed ratios that average 154.13.
• Joe Horn’s program that converts decimals to fractions has a speed ratio of 161.83.
Looking at the above results you can see that speed ratios of 150 and higher appear in register operations,
rolling down the stack, addition, subtraction, the INT function, the FRAC function, the program that
calculates e^ π (using the Taylor approximation), and the decimal to fraction program. The two programs
in the list show high speed ratios, perhaps, because they do not use functions that exhibit lower speed
ratios (like the trigonometric, hyperbolic, and logarithmic functions.) This conclusion agrees with the
relatively lower speed ratio of the statistical summation test. The listing for that test shows that it uses
two RAN# commands (which show speed ratios of about 113). The speed ratio for the statistical
summation test is very close to the speed ratio of the RAN# command. This observation tells me that the
RAN# is the slowest part of the statistical summation test—akin to a bottleneck, if you like.

The results of the speed comparison have eliminated my own initial naïve expectation that the new HP-
15C+ would have the exact same gain in speed across the board--for each and every operation. The
results show more gain in certain operations than others. Nevertheless, even in the lower speed ratios, we
are still talking about a new machine that is at least one hundred times faster than its predecessor. You
can say that HP put a tiger in the new CPU! If the late actor Jackie Gleason was a dedicated HP-15C fan
he would have said about the new machine, “How sweet it is!”
The HP-15C Limited Edition shows speed ratios, compared to the vintage
HP-15C, in the range of 102 to 189. Even at the lower end of this range, the
new HP-15C offers very impressive speed!
Table 2. Timing Results for the HP-15C

Adjusted
Test Raw Time Shift Time Loops Result
(seconds) (seconds) (seconds) Ops/sec
Empty loop 100 iterations 32.705 0 32.705 100 3.06
Empty loop 10 iterations 3.89 0 3.89 10 2.57
STO 9 39.93 32.705 7.225 100 13.84
RCL 9 40.71 32.705 8.005 100 12.49
RCL 9, RCL 8 for 100
iterations 48.83 32.705 16.125 100 6.20
RCL 9, RCL 8 for 10
iterations 5.5 3.89 1.61 10 6.21
Roll entire stack down 60.6 32.705 27.895 100 3.58
+ 42.26 32.705 9.555 100 10.47
- 42.21 32.705 9.505 100 10.52
* 44.36 32.705 11.655 100 8.58
/ 47.77 32.705 15.065 100 6.64
997, SQRT 55.9 40.71 15.19 100 6.58
997, X^2 52.32 40.71 11.61 100 8.61
π, 1/X 55.11 40.71 14.4 100 6.94
π, π, Y^X 151.5 48.83 102.67 100 0.97
SIN(0.6 RAD) 125.19 40.71 84.48 100 1.18
COS(0.6 RAD) 120.77 40.71 80.06 100 1.25
TAN(0.6 RAD) 93.53 40.71 52.82 100 1.89
ASIN(0.6) 108.14 40.71 67.43 100 1.48
ACOS(0.6) 111.4 40.71 70.69 100 1.41
ATAN(0.6) 84.89 40.71 44.18 100 2.26
SINH(0.6) 101.94 40.71 61.23 100 1.63
COSH(0.6) 91.03 40.71 50.32 180 3.58
TANH(.6) 121.59 40.71 80.88 100 1.24
ASINH(π) 125.64 40.71 84.93 100 1.18
ACOSH(π) 120.46 40.71 79.75 100 1.25
ATANH(0.6) 83.78 40.71 43.07 100 2.32

Adjusted
45, 45, ->P 89.69 48.83 40.86 100 2.45
π/4, 63,->R 152.51 48.83 103.68 100 0.96
1.3333, ->H 51.3 40.71 10.59 100 9.44
1.5591666667,->HMS 49.35 40.71 8.64 100 11.57
LN(997) 68.07 40.71 27.36 100 3.65
LOG(997) 77.18 40.71 36.47 100 2.74
EXP(2.123) 78.23 40.71 37.52 100 2.67
10^2.123 77.94 40.71 37.23 100 2.69
π, X! 237.53 40.71 196.82 100 0.51
Random number 45.88 32.705 13.175 100 7.59
2 Random numbers 59.3 32.705 26.595 100 3.76
FRAC of π 48.27 40.71 7.56 100 13.23
INT of π 48.06 40.71 7.35 100 13.61
Integrate 1/x from 1 to 100 334.97 5.5 329.47 10 3.04E-02
guesses 3 and 4 124.16 5.5 118.66 10 8.43E-02
Adding pairs of random
numbers in stat summations 115.56 59.3 56.26 100 1.78
Calculate e^π using FIX 7 270.83 3.89 266.94 10 3.75E-02
Calculate e^π using FIX 4 214.07 3.89 210.18 10 4.76E-02
Fractions for golden ratio in
FIX 9 660.23 3.89 656.34 10 1.52E-02
Table 3. Timing results for the HP-15C+

Adjusted
Empty loop 10000
iterations 21.1 0 21.1 10000 473.93
Empty loop 1000
iterations 2.51 0 2.51 1000 398.41
STO 9 25.14 21.1 4.04 10000 2475.25
RCL 9 25.54 21.1 4.44 10000 2252.25
iterations 30.03 21.1 8.93 10000 1119.82
iterations 3.36 2.51 0.85 1000 1176.47
Roll entire stack down 37.9 21.1 16.8 10000 595.24
+ 27.46 21.1 6.36 10000 1572.33
- 27.1 21.1 6 10000 1666.67
* 31.92 21.1 10.82 10000 924.21
/ 34.43 21.1 13.33 10000 750.19

Adjusted
997, SQRT 39.16 25.54 13.62 10000 734.21
997, X^2 34.53 25.54 8.99 10000 1112.35
π, 1/X 38.58 25.54 13.04 10000 766.87
π, π, Y^X 124.5 30.03 94.47 10000 105.85
SIN(0.6 RAD) 106.63 25.54 81.09 10000 123.32
COS(0.6 RAD) 100.7 25.54 75.16 10000 133.05
TAN(0.6 RAD) 72.8 25.54 47.26 10000 211.60
ASIN(0.6) 83.33 25.54 57.79 10000 173.04
ACOS(0.6) 86.57 25.54 61.03 10000 163.85
ATAN(0.6) 63.55 25.54 38.01 10000 263.09
SINH(0.6) 80.28 25.54 54.74 10000 182.68
COSH(0.6) 69.52 25.54 43.98 10000 227.38
TANH(.6) 101.67 25.54 76.13 10000 131.35
ASINH(π) 107.2 25.54 81.66 10000 122.46
ACOSH(π) 100.83 25.54 75.29 10000 132.82
ATANH(0.6) 60.74 25.54 35.2 10000 284.09
45, 45, ->P 63.01 30.03 32.98 10000 303.21
π/4, 63,->R 131.2 30.03 101.17 10000 98.84
1.3333, ->H 32.62 25.54 7.08 10000 1412.43
1.5591666667,->HMS 31.08 25.54 5.54 10000 1805.05
LN(997) 48.99 25.54 23.45 10000 426.44
LOG(997) 59.95 25.54 34.41 10000 290.61
EXP(2.123) 57.92 25.54 32.38 10000 308.83
10^2.123 58.26 25.54 32.72 10000 305.62
π, X! 194.44 25.54 168.9 10000 59.21
Random number 32.73 21.1 11.63 10000 859.85
2 Random numbers 44.34 21.1 23.24 10000 430.29
FRAC of π 30.21 25.54 4.67 10000 2141.33
INT of π 30.03 25.54 4.49 10000 2227.17
Integrate 1/x from 1 to
100 277.73 2.51 275.22 1000 3.63
guesses 3 and 4 93.81 2.51 91.3 1000 10.95
Adding pairs of random
numbers in stat
summations 92.94 44.34 48.6 10000 205.76
Calculate e^π using FIX
7 176.48 2.51 173.97 1000 5.75
Calculate e^π using FIX
4 138.26 2.51 135.75 1000 7.37
Fractions for golden ratio
in FIX 9 408.09 2.51 405.58 1000 2.47

References
1. HP-15C Owner's Handbook, June 1987, Hewlett-Packard.
2. HP-15C Advanced Functions Handbook, August 1982, Hewlett-Packard.
About the Author

Namir Shammas is a native of Baghdad, Iraq. He resides in Richmond, Virginia, USA.
Namir graduated with a degree in Chemical Engineering. He received a master degree
in Chemical engineering from the University of Michigan, Ann Arbor. He worked for a
few years in the field of water treatment before focusing for 17 years on writing
programming books and articles. Later he worked in corporate technical document-
ation. He is a big fan of HP calculators and collects many vintage models. His hobbies
also include traveling, music, movies (especially French movies), chemistry,
cosmology, Jungian psychology, mythology, statistics, and math. As a former PPC and
CHHU member, Namir enjoys attending the HHC conferences. Email him at:
nshammas@aol.com

How Large is 10^99?

U UH HU U
Return to Top
How Large is 1099?
Richard J. Nelson
Introduction
The dynamic range of many scientific calculators (entry level) is specified as -9.999999999 x 10-99 for the
smallest number to 9.999999999 x 1099 for the largest number if it is a ten digit machine. This was the
range of the first scientific calculator the 1972 HP-35A. How large or small is the number represented by
10±99?
As a number 1099 is 1 followed by 99 zeros. 1099 = 1,000,000,000,000,000,000,000,000,000,000,000,000,
000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000.
Add another zero and the number has a special name, googol (not Googel). Ten raised to the power of
googol is called a googolplex. This number is impossible to write out.
In the early 1970’s scientific calculators were often called electronic slide rules and this 10±99 range was
simply astounding. It was so unbelievable that the literature of the time often explained that all known
values in the universe would easily be covered by numbers within the 10±99range. Certainly all known
physical constants were not known beyond six or seven significant figures so the HP-35A was what
would be called today “The Killer Ap” of its day. Let’s explore a few real world “numbers” just to see
how meaningful 10±99 really is.
Students of my generation could make astounding calculations

in their heads because they used a slide rule for their daily
numerical problem solving(1). They had to do this because the
slide rule could not keep track of the magnitude of the “answer.” Fig. 1 – Typical slide rule of the 1960’s.
An answer might be 2.75, 0.000275, or 2.75 x 1013. Two problems had to be solved, one for the
significant digits and one the power of ten of the answer. The power of ten part was done in your head.
The student quickly learned to express all problems with the numbers in scientific notation. Let’s use the
example of the problem of multiplying 321 x 57,943.
Mentally the problem is expressed as 3 x 102 multiplied by 6 x 104. 3 x 6 = 18. You add exponents if
they are multiplied so the answer is 18 x 106 or in normal form 1.8 x 107. The correct answer from my
calculator is 18,599,703 or 1.8 x 107 as an approximation. Teachers often lament that students all too
often just mindlessly accept the answer they see in the calculator display without their having any real
world sense for the number. This is the reason you should always “sketch out the problem” just as a
technician or engineer must always sketch out a circuit he or she is wiring no matter how simple it may be
– even a battery, switch, resistor, and LED series circuit. To do otherwise is just plain sloppy work (prone
to human error) no matter how brilliant you may be.
Example 1: How big is the universe?

The idea here is to find something really big. Let’s calculate the volume, in cubic inches, of the
(theoretically) known universe which is, by the current Big Bang cosmology(2) theory, estimated to have a
radius of about 45.7 billion light years. In scientific notation this is 4.57 x 1010 light years. Assume that
the universe is spherical in shape. The volume of a sphere = 4πr3/3.
A light year is the distance light travels in one year. Since we need to calculate the answer in cubic inches

we need to know the distance in inches that light travels in one year. I will use the HP48/49/50 type of
calculator to solve this problem because it has all the constants I need built in.
a. The speed of light is: 186,282.397051 miles per

second (1.86282397051 x 105).
b. The length of a year is: 365.2421198781 days or

31,556,925.9747 seconds. (3.652421198781 x
102 or 3.155556,925.9747 x 107 seconds)
c. Distance light travels in one light year:

186,282.397051 mi/s x 31,556,925.9747 s =
5.87849981413 x 1012 miles per year.
d. Distance light travels in one year in feet:

5.87849981413 x 1012 mi. x 5,280 ft/mi =
3.10384790186 x 1016 feet per year.
e. Distance light travels in one year in inches:

3.10384790186 x 1016 x 12 in/ft =
3.72461748223 x 1017 inches per year.
The Universe volume in cubic inches = Fig. 2 – Universe

(1)
model showing estimated dimensions.
4*π(3.72461748223 x 1017)3/3
= 2.16438064428 x 1053 cubic inches.
The distance light travels per year in mm = 3.72461748223 x 1017 in/lt yr. x 25.4 mm/in =
9.4605284052 x 1018 mm/lt yr.
The Universe volume in cubic millimeters = 4*π(9.4605284052 x 1018)3/3 = 3.54678441413 x 1057 mm3.
Only 1057? That number is quite small compared to 1099.

Example 2: How many water molecules could there be in the Universe?
Let’s calculate how many atoms there are in a 45.7 billion light year Universe
that is completely composed of water - neglecting black holes, etc.
The number of atoms of ANY substance in a volume is: # of atoms = N*d*v/

(Molecular Weight).
Where: N = Avogadro’s number = 6.0221367 x 1023 atoms/mole. Fig. 3 – Earth is covered

d = Density = 1g/cc. 71% with water.
v = volume in cc.
From your chemistry classes you may remember that a mole of any substance has exactly the same
number of atoms (or molecules) no matter what the substance is made of. Water is made up of 2
hydrogen atoms and one oxygen atom. Hydrogen has an atomic weight of 1 and oxygen has an atomic
weight of 16. Water, H2O, has a molecular weight of 18.
A cubic centimeter of water is 1 gram. The number of moles is 1/18 = 5.55555555556 x 10-2 moles.

The number of molecules is therefore:
6.0221367 x 1023 * 5.55555555556 x 10-2 = 3.3456315 x 1022 molecules per cc. The number of atoms is
3 times larger because each molecule has three atoms, so there are 1.00368945 x 1023 atoms of water per
cubic centimeter, cc, of water.
The answer from example 1 calculated the universe as 3.54678441413 x 1057 mm3. This value converted
to cm3 is divided by 1,000 mm3/cm3 which = 3.54678441413 x 1054 cm3 in the Universe.
The number of water molecules to fill the Universe is 3.54678441413 x 1054 cm3 x 1.00368945 x 1023
molecules /cm3.= 3.55987009789 x 1077 H2O molecules.
Only 1077? That number is still quite small compared to 1099.
Exception I am limiting my search for a large number to real world

values specifically omitting “calculated” mathematics values such
as the number of piano pieces that could be written using each of the
88 keys on the piano once = 88! = 1.85482642257 x 10134. See Fig
4 and Fig 5 for HP-35s keys that may be used to calculate functions Fig. 4 – Factorial key. Fig. 5 – Stat key.
that result in very large numbers. Another exception for our reality number search is Graham’s number(3)
which cannot be expressed using the conventional notation of powers, and powers of powers..
Example 3: How old is the Universe in nanoseconds?
The age of the Universe is usually based on the Hubble constant which is a measure of the current expan-
sion rate of the universe. Many astronomers are working to measure the Hubble constant using different
techniques. Until recently, the best estimates ranged from 65 km/sec/Megaparsec to 80
km/sec/Megaparsec, with the best value being about 72 km/sec/Megaparsec. In more familiar units,
astronomers believe that age of the Universe is between 12 and 14 billion years. Let’s use 14 billion
years.
From example 1.
b. The length of a year is: 365.2421198781 days or 31,556,925.9747 seconds.
14 billion is 1.4 x 1010 yrs x 3.1556,9259747 x 107sec./yr. = 4.41796963646 x 1017 = Universe age in
seconds.
If the big bang theory theorizes the early events in one nanosecond increments how old is the Universe in
nanoseconds (light travels about one foot in one nanosecond)? There are 109 nanoseconds in one second.
The age of the Universe in nanoseconds is approximately 4.41796963646 x 1026 nanoseconds.
Only 1026? That number is almost tiny compared to 1099.
Can the reader think of any physical reality object that requires its value to be expressed as a numeric
value that cannot be calculated by an HP scientific or graphing calculator?
Obversations and Conclusions
How Large is 1099?

Our largest samplings of “real world” numbers(4) are:
1. The Universe volume in cubic millimeters = 3.5 x 1057. (3.54678441413 x 1057).
2. The number of water molecules to fill the Universe = 3.6 x 1077 (3.55987009789 x 1077).
3. The age of the Universe in nanoseconds is approximately 4.4 x 1026. (4.41796963646 x 1026 ns.)
These numbers are quite small compared to 1099.

Even if we assume that we have missed a few numbers by a factor of a thousand (103) or ten thousand
(104) none of our real world examples come close to 1099.
Our current generation of calculators have an even wider dynamic range with the upper limit of 10±499.
As mentioned with our piano key example it is easy to generate very large or very small numbers
mathematically. The new question is, “How Large is 10499?”
It is certainly larger than any reality numbers you may think about. Any factorial calculation greater than
253 (253! = 5.17346099264 x 10499) will cause overflow, but when it comes to “pure” mathematics a
finite state machine like a scientific calculator is essentially useless. The important question to ask is:
What are realistic dynamic range limits for a scientific or graphing calculator?
Epilog
Modern graphing calculators have morphed into symbolic math machines. The student asks: Is the hand
held calculator adequate?
This topic was examined by Wlodek Mier-Jedrzejowicz in a series of talks at

titled “Would Einstein Have Used One?”, “Could Einstein have Used One?”, and “Should Einstein Have
Used One? The whole range of mathematics is addressed with a review of what advanced mathematics is
calculator solvable today and what could or should be added in the future. From the 2009 presentation;
“In the second presentation I asked if Einstein could have done some General Relativity work on an HP50g, even if
this were to be very difficult. Say he was shipwrecked on a deserted island with only an HP50g for company, could
he have used it? Fundamentally, the Computer Algebra System on an HP50g is just about powerful enough to do
some symbolic tensor calculus, but I doubt that Einstein or anyone else would happily do so. Instead, I suggested
that if any General Relativity were to be done on an HP50g, the programming would have to be done first in some
other language than RPL. If Einstein were to use an HP50g he could do so, but he would have to learn to program
in an appropriate language, as well as learn to use Tensor Calculus, which took him some effort. He would need
some other computer to write the programs first, then translate them into code to run on the HP50g, and transfer it
to the HP50g. So he could just about have used an HP50g on his desert island, but with considerable difficulty.”
Finishing up Wlodek said;
“This means that if E wants to work with tensors on an HP calculator, then he or she has to get someone to write
the programs for them. This was discussed on comp.sys.hp48 two years ago, but at that time no-one had done it
yet. I just wonder if anyone will ever have the time or the wish to do so. And that is the answer to “Should E use
one?” They won’t know until they try! So, they should find someone willing to write the necessary programs for an
HP50g! I’m not volunteering unless I get a substantial university grant to do it. If I do, I’ll let you know next year.
Otherwise I might go back to programming on my HP-41 and give a talk on that.

I’d like to finish by thanking everyone who has made suggestions for these presentations and at them – especially
Roger Hill, Richard Nelson and Bill Butler who pointed out to me, just as I was finishing writing this, that HP50g
tensor software is now available from http://www.heuson-software.de/ “
____________________________________________________________________________________
Notes
(1) See HP Solve issue 21 for an article, Calculating before Calculators, that describes the various pre-calculator
methods of performing numerical calculations.
http://h20331.www2.hp.com/hpsub/downloads/Newsletters_HP_Calculator_eNL_12_Dec_2010_v1.pdf
(2) This number is from: http://en.wikipedia.org/wiki/Observable_universe .
(3) See http://www-users.cs.york.ac.uk/susan/cyc/g/graham.htm for an explanation of Graham’s number and other
large numbers and how they are named and expressed.
(4) All of the numbers given in this article are silly in terms of the number of digits provided. They are given as
shown in your 12 digit calculator display as a means of verifying that you have pressed the correct keys and
you have the correct answer.
About the Author

Richard J. Nelson is a long time HP Calculator enthusiast. He was editor and
publisher of HP-65 Notes, The PPC Journal, The PPC Calculator Journal, and
the CHHU Chronicle. He has also had articles published in HP65 Key Note
and HP Key Notes. As an Electronics Engineer turned technical writer Richard
has published hundreds of articles discussing all aspects of HP Calculators.
His work may be found on the Internet and the HCC websites at: hhuc.us He
proposed and published the PPC ROM and actively contributed to the UK
HPCC book, RCL 20. His primary calculator interest is the User Interface.

From the Editor

U UH HU U
Return to Top
From The Editor – Issue 25
Fall is a time for the HP Handheld Conference, and active and serious HP calculator users have been
registering in exceptional numbers to attend HHC 2011 in San Diego the last weekend in September.
Since this issue of HP Solve is a bit early for the obvious reasons of new product announcements a
Conference report will have to wait for the next issue, #26.
HP News
HP hosted an HP
August 30, 2011, NYC @ Harry’s Cafe & Steak.
th
12c 30 Anniversary Celebration for the financial
press. This event also served to announce the re-
release of the popular HP-15C now known as the
HP 15c Limited Edition.
Because so many of the press are “electronic” these

days a two day lead was considered adequate for
the official announcement date of September 1,
2011 for both machines. Laura Harich, HP
Calculator Marketing (CA) and Mike Hockey HP PR
(TX) provided support for the event. Gene Wright, Photo by Jake Schwartz
nd
well known financial, author and HP Solve Fig. 1 – Lower left, 2 from left HPs Laura Harich.
nd
contributor (TN) attended as did Jake Schwartz Lower right 2 from front HP 12c author Gene Wright.
another well known HP Solve contributor (NJ). See photo above. Aside from answering questions and
providing personal background information the group viewed an HP video which covered the long history
of the HP 12c including an interview with one of its developers, Dennis Harms.
There was even a business card drawing for one of each of the two new products. Radiris Diaz of
CuteGeek.com won the HP 12c 30th Anniversary Edition and Stacy Cowley of CNNMoney.com won the
HP 15c Limited Edition.
The press event was planned in advance of the storm, but everything dried out just in time for everyone to
have a great evening sharing stories of how the finance world has been impacted by the over 30 years of
HP-12C usage.
Community News
As I mentioned in the last issue, the HP user Community is participating in the “designing” of two new
calculators. If you think that it is an easy task to design a calculator you soon realize that the devil is in
the details. One of the most challenging aspects is accuracy - a hallmark of HP calculators. Another
aspect is the documentation.
Suppose you are King and you make all the decisions. You have the factory standing by to start the new
model on a specific date. If you don’t release the new design the factory will (instead) make another
model (HP-12C’s are always needed ) and the new model production will have to be moved to another
date – many months later. Now you have to make an important change at the last minute. What do you
do? Do you delay the machine and miss an important sales period such as Christmas or back to school?
In some situations the change could ripple through several teams (to change the hardware and/or the
documentation, etc.) to extend the introduction a full year. The market radically changes in one year.
Another topic being presented at HHC 2011 is a new HP calculator restoration book by Geoff Quickfall.
HP Solve readers will remember that Geoff was interviewed in HP Solve issue 12, and the book was

previewed in issue 19, page 23. Geoff is also a Beta HP-41CL user/documentator so he is busy indeed -
and he will have a lot to talk about at HHC 2011. hhuc.us
Another very exciting project repurposes one of HP’s business calculators (HP 20b or HP 30b) to create a
new calculator dubbed the WP 34s as mentioned in this column in issue 24. The WP 34s calculator
continues to develop with features added nearly every week. Originally it had only 506 program steps,
but through very clever firmware programming, it now has 8 flash banks of 506 steps that can hold user
callable programs, utilities, or subroutines. This is quite capable indeed. One of the areas not addressed
by built-in functionality of the WP 34s is matrix math. Jake Schwartz and Gene Wright filled in this
functionality by adapting the 30 year old PPC ROM matrix subroutines M1 through M5 (and a couple of
others) and the PPC ROM application program RRM and now the WP 34s has a matrix handling program
that can solve a system of 6 equations, find the inverse of a 6x6 matrix, or the determinant of an 8x8
matrix. By the way, the WP 34s is fast. Thanks HP. A 7x7 determinant was computed in just under 2
seconds. Wow. There was one fundamental difference, however. The HP-41 SIGN function was
“defined” differently, and they had to add two steps for each time it was used in the routines. The HP-41
returns -1, +1, & +1 if the number is -, 0, or +. The WP 34s returns -1, 0, +1 under the same conditions.
If you were King what decision would you make? It makes you wonder what the HP-41 developers were
thinking. Was there some hardware restriction?
Here is the content of this issue.
S01 – The HP-12C is Announced This one of a kind business calculator has the
unique distinction of being manufactured longer than any other calculator – July 4 1980 to present
inclusive is 32 years. See additional article related to the HP-12C.
S02 – The HP-15C Limited Edition, LE, is Announced. This is a re-release of the famous HP-15C,
one of (now, including the 12C 30th anniversary machine ) nine machines in the Voyager Series of HP
calculators. The 15C LE enjoys the exceptional accuracy of the improved algorithms of the Voyager
series.
S03 – The HP-12C, 30 Years and Counting by Gene Wright and me. This is a short review of the very
popular business calculator with new information related to the manufacturing history and the MSRP
pricing over the 32 years of its being manufacturered.
S04 – Calculator Programmability – How Important is it in 2011? Is there an advantage to having a

scientific calculator programmable? What are the reasons for programmability? This article examines
the subject and concludes with: The large number of features, memory size, and super speed of the HP 15
LE makes its programmability an important feature when compared to other calculators - for the 12
programming reasons listed.
S05 – Converting Decimal Numbers to Fractions by Jospeh Horn. Joseph explains the continued
Fraction Algorithm and provides an RPN HP-15C program that quickly converts any decimal to a fraction
to the accuracy set by the FIX display value. The program listing provides program line numbers for
easier conversion to other RPN calculators.
S06 – Limited Edition HP-15C Execution Times by Namir Shammas. Namir examined the new
incarnation of the 15C LE and compared the execution times for many of the instructions, programs, and
built-in applications with those of the original 15C. The results were surprising because the expectation
was that it would be faster by some factor. This turned out to be very untrue. The speed advantage
actually ranged from 102 to 180 times faster. A total of 44 speed ratios were measured, the total is
5,659.06 and the average is 128.615. Saying that that the HP-15C LI is about 128 times faster is a good
average value to use.

S07 – How Large is 1099? Ten digit calculators usually have an exponent range of ±99. Exactly how
large is this number and how does it compare with the real universe? Three examples of the largest
possible physical values are used to see if 10±99 is big or small enough to calculate numbers in the real
world. Twelve digit machines provide an exponent of 10±499. How important is this limit?
S08 – Regular Columns This is a collection of news items and repeating/regular columns. A new
column, Calculator Accuracy, starts with this issue.
♦ From the editor. This column provides feedback and commentary from the editor.
♦ One Minute Marvels. This OMM provides a routine to determine when a Friday the 13th occurs.
The incredible calendar built into the 48/49/50 series of machines makes this a very easy task. Tables
are provided for the number of F13’s from 1600 to 3299. Details on how this was done on request.
♦ Calculator Accuracy. This is an important topic for HP calculator users to explore and use to be
better calculator users. No calculator is 100% accurate – yet.
S09 – #8 in the Math Review Series: Mathematical Constants – π This constant is probably one of
the most well-known of all mathematical constants. Here is an overview of this interesting number
with lots of links for further astonishment and study.
That is it for this issue. I hope you enjoy it. If not, tell me!
Also tell me what you liked, and what you would like to read about.
X < > Y,
Richard
Email me at: hpsolve@hp.com
HP 48 One Minute Marvel – No. 12 – Friday the 13th

One Minute Marvels, OMMs, are short, efficient, unusual, and fun HP 48 programs that may be entered
into your machine in a minute or less. These programs were developed on the HP 48, but they will
usually run on the HP 49 and HP 50 as well. Note the HP48 byte count is for the program only.
One of the really powerful functions of the HP-48/49/50 series of RPL machine is the built in 8,419 year
(October 15, 1582 to December 31, 9999) calendar. Using the powerful date commands all kinds of
powerful date routines may be written. Some people enjoy the fun of a special date in the United States
(other countries have similar days and dates, in Columbia and Spain it is Tuesday the 13th) when Friday
falls on the 13th of the month.
Every year has at least one Friday the 13th. What is the maximum number possible in one year? The
following 14 years from 1900 to 2000 have three Friday the 13ths: 1903, 1914, 1925, 1928, 1931, 1942,
1953, 1956, 1959, 1970, 1981, 1984, 1987, 1998. Now that we are well into the 21st century you may
easily determine how many years will have three Friday the 13ths in this century.
Friday the 13th (Joseph K. Horn)
Key in a year and ‘FRI13’ will return the dates of Fridays that occur on the 13th of the month.
‘FRI13’ << 1000000 / 1.13 + 13 FOR d 6.133 d DDAYS 7 MOD NOT d IFT NEXT >>
16 commands, 88.0 Bytes, # 670Bh. Timing: 1987⇒ 2.131987, 3.131987, 11.131987 in 0.148_sec.

What is the answer to the question of the occurrences of Friday The 13th for various centuries? I made the
tables below using ‘FRI13’. See the observations below the tables.
Table 1 — 17th Through 33rd Century (traditional) Table 2 — 17th Through 33rd Century (popular)
th
Friday The 13 Occurrences Friday The 13th Occurrences
Occurrences Per Year Occurrences Per Year
# Century Period Total Once Twice Thrice Century Period Total Once Twice Thrice
1. 17th 1601 - 1700 172 43 42 15 17th 1600 - 1699 172 43 42 15
2. 18th 1701 - 1800 172 43 42 15 18th 1700 - 1799 172 43 42 15
3. 19th 1801 - 1900 173 42 43 15 19th 1800 - 1899 172 43 42 15
th
4. 20 1901 - 2000 171 43 43 14 20th 1900 - 1999 172 42 44 14
5. 21st 2001 - 2100 172 43 42 15 21st 2000 - 2099 172 43 42 15
6. 22nd 2101 - 2100 172 43 42 15 22nd 2100 - 2199 172 43 42 15
7. 23rd 2201 - 2200 173 42 43 15 23rd 2200 - 2299 172 43 42 15
th
8. 24 2301 - 2300 171 43 43 14 24th 2300 - 2399 172 42 44 14
9. 25th 2401 - 2400 172 43 42 15 25th 2400 - 2499 172 43 42 15
10. 26th 2501 - 2500 172 43 42 15 26th 2500 - 2599 172 43 42 15
th
11. 27 2601 - 2600 173 42 43 15 27th 2600 - 2699 172 43 42 15
12. 28th 2701 - 2700 171 43 43 14 28th 2700 - 2799 172 42 44 14
13. 29th 2801 - 2800 172 43 42 15 29th 2800 - 2899 172 43 42 15
14. 30th 2901 - 2009 172 43 42 15 30th 2900 - 2999 172 43 42 15
st
15. 31 3001 - 3000 173 42 43 15 31st 3000 - 3099 172 43 42 15
16. 32nd 3101 - 3100 171 43 43 14 32nd 3100 - 3199 172 42 44 14
17. 33rd 3201 - 3200 172 43 42 15 33rd 3200 - 3299 172 43 42 15
Observations:
1. Occurrences per century. Depending on how you wish to define a century - I called one
traditional and the other popular - the total number of occurrences of F13 per century vary. The
popular belief of the recent new millennium starting on January 1st 2,000 picks up the cycle “out of
sync” so that there are not 172 Friday The 13ths for each century.
2. The calendar repeats every 4 centuries. Roger Hill participated in a contest to print calendars and
he investigated them. I asked him for his thoughts. His comment is: "One thing to note is that, for
the presently-used Gregorian calendar, you only have to look at a 400-year interval, say 1600-1999,
because the calendar repeats every 400 years." The total number of Friday The 13ths per cycle is
688.
3. Having three Friday The 13ths in one year occurs 15% of the time. Roger observes: "I did some
research on Friday the 13ths (or F13's for short) a long time ago and (I think) figured out the
probability of having one, two or three in a year chosen at random. I don't remember what I got
now, but with a little luck I might be able to find it."
"Another thing is that, in any year, the 7 months from May through November (inclusive) always
start on 7 different days, so there will always be one and only one F13 in that 7-month interval. For
non-leap years, the ones that have three F13's are the ones in which February 13 and March 13 are
on a Friday, which happens in about 1 year out of 7. The criterion is different for a leap year, but
again it happens in about 1 leap year out of 7. Hence the 14 or 15 per century that you found." I did
not pick up that bit of information from the tables.
4. Having one or two Friday The 13ths in one year is about equal probability. This is obvious
from the tables. Roger notes: "An interesting fact about the Gregorian calendar is that, while you
would expect a year chosen at random to be equally likely to start on any day of the week, over the

400 year period of repetition it turns out that some days of the week are very slightly favored over
others. For example, the probability that a year chosen at random starts on a Sunday is very close to
1/7 but not exactly. I've forgotten the exact results, but (again with a little luck) I could probably
find them again. (It was almost 30 years ago when I fooled around with all this...)
Calculator Accuracy – Part 1 - Calculator Forensics

Introduction
When we pick up a calculator and make a calculation we place a great deal of faith that the numbers
resulting from the calculation are correct. Hewlett Packard created the first scientific calculator with the
HP-35A and as part of the specifications of that machine there was an accuracy statement. For the basic
math functions of +, -, x , ÷, and √ the answer is supposed to be within one count in the last place,
typically the 10th (or 12th) digit). Transcendental functions were slightly less accurate.
Function accuracy, however, is only one part of the accuracy “specification.” If the calculator has built in
applications such as complex numbers, solve, integrate, or matrices the accuracy specification is even
less. In some cases surprisingly, perhaps three digits, less. The average user will seldom consult the
User’s Guide for an accuracy statement – if there is one – to know how reliable, how accurate, all those
digits are. After all, the calculator designers should know what they are doing.
Customers needing to buy a calculator won’t spend very much time delving into the details of calculator
accuracy other than to hear about the reputation of a particular manufacturer.
The calculator programmers that microcode the internal math routines spend their time trying to reduce
the number of instructions and to optimize them to run as fast as possible. They cannot be expected to be
numerical analysts in terms of mathematical accuracy. They just implement the algorithms they are
given.
In order to determine the accuracy of the number in the display, with its limited fixed number of digits,
you must know the correct answer to several digits greater than that produced by the function. Casual
calculator users just want to know how the various models compare, and they have devised various “tests”
to compare machines.
Comparison testing
One such scientific model test has been proposed by Mike Sebastian. He has a website dedicated to what
he calls Calculator Forensics(1). Below is the “forensics” data for HP calculators. The results in Tables 1
and 2 are for his single forensics trigonometry function test. Some of the data is current data not found in
Mikes 2007 table(2). The tables also add a Rating column and an Error column for comparison purposes.
The models are arranged in increasing accuracy order.
Here is the test problem. In degrees mode calculate arcsin (arccos (arctan (tan (cos (sin (9) ) ) ) ) )
algebraically, and 9 SIN COS TAN ARCTAN ARCCOS ARCSIN in RPN. A rating is suggested for
comparison purposes with the higher the value the better. The idea is that the function-inverse is
supposed to return the input, in this case 9 degrees. Supposedly the more accurate calculator (overall?) is
the one that is closest to 9. All calculators in the tables use BCD arithmetic except as noted + in table 1.

Table 1 - Mike Sebastian's Calculator Forensics Results for 10D HP Models
Calculators Result Rating* Error
HP-35A with bug 9.002 983 113 2 2.98311x10-3
HP-45A, 65A, 9.004 076 644 2 4.07664 x10-3
HP-35A bug corrected 9.004 076 901 2 4.076901x10-3
HP-21, 25/C, 9.004 076 649 2 4.076649x10-3
HP-55A 9.004 076 898 2 4.076898 x10-3
HP-41C/Cv/CX, 10C, 11C, 15C, 27,
29C, 30E, 32E, 33E, 34C, 67A, 91A, 9.000 417 403 3 4.1743x10-4
97A/S
HP 9g+, 30s+ 9.000 000 000 9 0+
* Simply the number of zeros or nines following the decimal point. The higher the better. The rating
is one less than the exponent of the error.
+ Uses 80 bit microprocessor binary arithmetic which provides at least 23 decimal digits.
Table 2 - Mike Sebastian's Calculator Forensics Results for 12D HP Models

Calculators Result Rating* Error
HP-19BII, 20b. 20S, 22S, 27S, 28C/S,
30b, 32S, 32SII, 38G, 39G, 42S,
8.999 998 642 67 5 -1.35733x10-6
48G/GII/G+/GX, 48S, 49g/g+, 50g, 71B,
75C/D,
HP-6s/6s Solar 8.999 996 370 4 5 -3.6296x10-6
HP 33S, 35s 8.999 999 860 01 5 -1.3999x10-6
HP 9s 8.999 998 643 82 5 -1.35618x10-6
HP 10s 9.000 000 342 383 6 3.4239x10-7
WP 34s (V1444) 9.000 000 000 03 10 3x10-11
* Simply the number of zeros or nines following the decimal point. The higher the better.
The rating is one less than the exponent of the error.
Observations and conclusions

If you are familiar with HP’s calculator models you will notice that as you go down the table the model
numbers are more recent. This demonstrates that HP’s calculators keep getting more accurate with time.
Table one values (ten digit machines) are all slightly high. Table two values (12 digit machines) are “all”
(96%) slightly low. Another “obvious” conclusion you can make is that various series of models produce
the same results. The only calculators that return 9 are the 80 bit (lowest cost) machines.
Calculator Accuracy - Part 2

Inherent to calculator accuracy is the topic we will discuss in issue 26 – Guard Digits. The correct use of
guard digits is what sets HP calculators apart from the others.
____________________________________________________________________________________
Notes: Calculator Accuracy – Part 1
(1). Mike makes no claim as to any mathematical (analysis) basis for his Calculator Forensics problem he
uses to arrive at a calculator comparison number. The “9” for example was chosen to be a single digit
(for ease of keying in) far from zero which is always troublesome for trigonometry functions. Obviously
he wanted to give the inaccurate machines more leeway. His website is at:
http://www.rskey.org/~mwsebastian/miscprj/forensics.htm
(2). Thanks to Palmer Hanson, Joseph Horn, and Jake Schwartz for contributing data.

Fundamentals of Applied Math Series #8
HP Solve #25 page 63
HP 82240B IR Printer Introduction
 Previous Article
U UH
Return to Top
#8 in Fundamentals of Applied Math Series
Mathematical Constants – π
Richard J. Nelson
Introduction – What is a mathematical constant?

A mathematical constant is a special number, usually a real number that is especially interesting and
useful to mathematicians. Constants(1) arise in many different areas of mathematics and two especially
well known constants are Euler’s number e, and Pi, π. e was discussed in Math Review #6, HP Solve
issue 23.
π is well known and every high school student learns that π is

the ratio of the circumference of any circle to its diameter. Fig. 1
shows 24 digits of π. While π is a mathematical constant it is
also found in many of the formulas of Physics and Electronics.
Albert Einstein suggested that the course length of looping rivers,

from source to end, compared to a straight course has a ratio
which approaches π. Fig. 1 – HP-35s Display showing π.
Then there is the π joke. (Forwarded to me by Roger Hill Hill, PPC ROM Manual contributor, HP
calculator author and enthusiast, meet Roger at HHC 2011).
Question: How can you tell if a person is a mathematician, a physicist, an engineer, or a salesman?
Answer: Ask him what π is.
The mathematician will say, "It is the ratio of the circumference of a circle to the diameter".
The physicist will say, "It's 3.14".
The engineer will say, "It's about 3".
The salesman will say, "It's normally 3, but I can get it to you for 2.6".
Discovering π
π has a very long colorful and interesting history. Table 1 is a brief portion of the Chronological Table
taken from A History of π by Petr Beckmann, see page 196 shown as reference 1 of Appendix B.
Table 1 – The Early History of π
Date π Status
Circa 2000 BC Babylonians use π = 3 1/8.
Circa 2000 BC Egyptians use π = (16/9)2 = 3.1605.
th
12 century BC Chinese use π = 3.
Circa 500 BC I Kings vii, 23 implies π = 3.
Circa 440 BC Hippocrates of Chios squares the lune.
Circa 434 BC Anaxagoras attempts to square the circle.
Circa 430 BC Antiphon enunciates the principle of exhaustion.
Circa 420 BC Hippias discovers the quadratix.
Circa 335 BC Dinostratos uses the quadratix to square the circle.

Date π Status
nd
3 century BC Archimedes establishes 3 10/71 < π < 3 1/7 and π = 211875: 67441 = 3.14163.
2nd century AD Ptolemy uses π = 377/120 = 3.14166…
3rd century AD Chung Hing uses π = 3.16…
236 AD Liu Hui uses π = 157/50 = 3.14
5th century Tsu Chung-Chi establishes 3.1415926 < π < 3.1415927
The relationship of the circle circumference to its diameter has been known for about 4,000 years, but the
first use of the π symbol was by William Jones in 1706 because it was the first letter of the Greek word
for "perimeter" or "periphery.” The great mathematician Leonhard Euler used it in 1737 and it has been
accepted since. π is a lower case greek letter pronunced “pie” that is not capitalized if it begins a
sentence because the upper case greek letter, Π, is used for designating the product of a sequence. If C is
the circumference of the circle and D is the diameter of the circle, π is defined as shown below.
Measuring π
You may determine a desired number of digits of π in one of three ways. You may measure it, you may
calculate it, or you may look it up on the internet. I really wanted to make an attempt to measure π, but
the daily summer 107 degrees here in the Sonoran desert discouraged me. Instead, let’s perform a thought
measurement. Here are the materials I was going to use.
1. Three metal stakes (12” nails).
2. 210 feet of white polypropylene cord.
3. 110 feet of bailing wire.
4. 100 foot measuring tape.
5. Bicycle (front wheel and fork). I would have to borrow one.
6. Masking tape.
7. Carpenters marking pencil and ball point pen.
8. Large flat level smooth layout area at least 220 x 110 feet. (half of a 200 foot diameter circle.)
Here is the measurement procedure. The goal is to make all measurements to a resolution of 1/8 inches.
A. Remove the front fork from the bicycle.
B. Wrap a piece of masking tape around the tire and wheel and mark a line down the outside
perpendicular to the ground.
C. Attach one end of the wire to the frame near the wheel axle.
D. Move all of your materials to the layout area and pound in two stakes about 220 feet apart and
stretch the white polypropylene cord to serve as the diameter of the circle.
E. Pound in the third stake in the center of the white cord to mark the center of the circle.
F. Position the wheel 100 feet from the center (an assistant is most helpful here) and make a loop to
be placed over the center stake.
G. Holding the wheel with the wire tight (it should not stretch) rotate and align the mark on the rim
with the white cord.
H. Keeping the wire tight walk the wire around the semicircle keeping the wheel perpendicular to the
surface counting the number of times the tape comes in contact with the surface. When you are

near the other end of the white diameter cord you will usually have less than one full rotation left.
Finish that last rotation and place a second piece of tape on the wheel to indicate the last portion of
the rotation. Write down the number of full rotations, gather your materials and head home.
I. In the car port use a long straight edge and draw a line on the concrete with a cross line at one end.
J. Use your wheel and align the first tape mark with the cross line and roll the wheel down the line
one revolution and make another cross line. This is the second measurement (the first was the
radius of the circle at 100 feet). Repeat the rotational measurement for the length to the second
tape.
K. Calculate π.
You must account for the diameter of the center stake in your measurements.
Let’s assume the radius (wire length) is 100 feet or 1200 inches. The diameter is 2400 inches. After
doubling the thought wheel measured total length (measured circumference x revolutions + extra partial
revolution) the circumference is 7,539 3/4hs (7,539.75 inches). I realize that making these measurements
to a resolution of 1/8th inches takes great skill and is very difficult. Making the division we have π =
3.1415625.
The theoretical circumference with a 100.00000 feet diameter is 7,539.82236862 inches. Let’s assume an
error of ¼ inch too high for the circumference and ¼ inch too low for the diameter for a maximum error
on the high side. The calculated theoretical value for π is 7,540.07236862 divided by 2,400.25 =
3.14136959426. The difference (high error) with 12 digits of π is 0.00022305933. This means that we
can easily depend on three accurate decimal digits or 3.141 using our measurement method. If our work
is really practiced and careful we might hope for 3.1415.
CONCLUSION: Physically measuring to determine π

using normal measuring methods is not very productive.
Calculating π
Most students don’t even think of the numerical value of π because it is a keystroke
constant of most HP scientific and graphing calcualtors. This most popular and very
easily entered irrational and transcendental number is found on the HP 35s shifted COS
key as shown in Fig. 2.
All scientific and graphing calcualtors have π internally stored to a value much greater
Fig. 2 - HP35s π than what the display will show. This is required to perform the trig functions to their
specified accuracy. The number of digits of π stored in the HP-48/49/50 calculators, as a constant, is 31.
Here is a technique suggested by Joseph K. Horn to obtain the digits of π that are used by your HP
calculator for the sin function. Using RAD mode type in as many digits of π as you wish up to the limit
of the display. Do not round, use the truncated value. Suppose I only remember that π is 3.14159 and I
have an HP-15C handy. After typing in the 6 digits I remember (let’s pretend that the π key is broken) I
press the SIN key. The display will show: 0.000002654. Use FIX 9. Now multiply the number by
10000 to see 0.026535900. That is all the useful digits we can get, but we now have 12 accurate rounded
digits of π.

Calculators use more digits than they display for their calculations and these extra digits are called guard
digits. HP calculators discard the guard digits (other calculator manufacturers usually do not) when the
answer is calculated to give you a “what you see is what you get” result. To demonstrate this press the π
key to see 3.141582654. If the calculator saved the guard digits you could subtract 3 and multiply by 10
to see the next guard digit. For HP calculators this will be zero. Subtract 1 and multiple by 10 again.
Now the last two digits are zero.
For 10 digit calculators the guard digits are usually two. For 12 digit calculators there are typically three
guard digits.
Now let’s use the “SIN π Digits” technique using the HP 35s. Keying in the maximum number (12) of
truncated digits, 3.14159265358, and taking the SIN we see 0.000000000001. Multiplying this by 1012
we will see 9.79323846264. We now have 24 accurate digits of π if we append these digits to our 12
input digits. This is shown in Fig. 1. No matter what you do you can’t get more digits. Checking the
normal π key for guard digits (the only other technique we have to squeeze out extra digits of a
calculation) all we get are 12 digits, 3.14159265359 which is the correctly rounded value of π to 12 digits.
Exactly what this means is unclear. We know that the Tangent function is the most sensitive to errors at
the extremes and that π is used internally by HP for the Tangent function to 31 digits in the later (1990)
RPL calculators. Below is a table of the current machines with a few historical models for reference.
Table 2 – Digits of π That May be Extracted From Your HP Calculator

Sin π Digits (RAD) HP Model(s) [π is given below for reference]
12 14 24 28
3.14159 26535 89793 23846 26433 83279 50288 41971 69399
12 HP 9s, 10C, 10s, 11C, 15C, 15C Limited Edition,
41C/CV/CX
14 HP 300s
24 HP 9g*, 20b, 30b, 33s, 35s, 39gs. 40gs, 48/49/50,
28 HP 20b/30b repurposed as WP-34s
* = Garbage digits after the indicated number of correct digits obtained by
looking at the guard digits.
Palmer Hanson offers an explanation on how the technique works.

Using the old trigonometry identity: sin(A + B) = sin A cos B + cos B sin A
For our problem A = π, and B = a small angle
Then sin(π + B) = sin π cos B + cos π sin B
and since sin π = 0 and cos π = -1 (in radian mode)
sin(pi + B) = -sin B
and for small B, good enough for what we are doing here
sin(pi + B) = -B
For pi + B = 3.141492653, (10D truncated value of π) B is negative so sin(π + B) = +B
For pi + B = 3.141592654, (10D rounded value of π) B is positive so sin(pi + B) = -B

This illustrates why it is best (simpler) to use a truncated input for Sin and to view the result in scientific
notation or multiply by E10 or E12 depending on the number of display digits. The result will then be as
shown in fig. 1.
π has been calculated to more than 5 trillion digits and memorized to 67,890 digits(5) by Lu Chao a 24-
year old chinese graduate student as recognized by the Guinness book of records. It took 24 hours and 4
minutes to speak them – without error! See Exploring π below to learn how this is done.
Calculating the decimal digits of π is a popular HP calculator programming challenge and various authors
have written programs to calculate the first 1,000 digits of π for the calculators of the 70’s and 80’s.
Often the various machines were compared for speed by using the number of hours it would take to
produce the first 1,000 π digits. This number was used because of memory restrictions. RPL calculators
of the 90’s and later have math libraries that calculate hugh numbers of the digits of π.
Calculating an approximation for π may be also be done by converting the decimal value desired to a
fraction using Joseph’s program found elsewhere in this issue. Using his program running on the HP 35s
results in the values shown in Table 3. The input is 3.14159265359. The accuracy is set by the FIX mode
which is one less than the Digits column.
Table 3 – Fractions That Evaluate to π

Desired Digit Nume- Denom
Accuracy s rator -inator Error
3 1 3 1 -0.14159265358
3.1 2 22 7 0.00126448927
3.14 3 22 7 0.00126448927
3.142 4 333 106 0.00008321963
3.1416 5 355 113 0.00000026676
3.14159 6 355 113 0.00000026676
3.141593 7 355 113 0.00000026676
3.1415927 8 103,993 33,102 -0.00000000058
3.14159265 9 103,993 33,102 -0.00000000058
3.1415922654 10 104348 33,215 0.00000000033
3.1415926536 11 1,146,408 364,913 0.00000000003
3.14159265359 12 1,146,408 364,913 0
If you use the program you will have two choices for the 12 digits. You may use a truncted input with the
twelfth digit 8 or you may use a rounded input with the twelfth digit 9. Joseph says the the rounded value
must be used because that is the mathematical basis for the continued fractions algorithm. If you use
3.14159265358 as the input several fractions will change and the error will be off by one digit.
Mathematicians have been calculating (approximating) π by ratios since the beginning and Beckmann
provides a table of these values as shown in Table 4.

Table 4 – Historical Fractions That Evaluate to π
Partial N & D values from Beckmann, page 171, Fig. A1
.
3 14159 26535 89793 23846 26433 832
12D Divided
N D Value Error
3 1 3.00000000000 -0.14159265359
22 7 3.14285714286 0.00126448927
333 106 3.14150943396 -0.00008321963
355 113 3.14159292035 0.00000026676
103,993 33,102 3.14158265301 -0.00000000058
104,348 33,215 3.14159265392 0.00000000033
208,341 66,317 3.14159265347 0.00000000012
312,689 99,532 3.14159265362 0.00000000003
833,719 26,5381 3.14159265358 0.00000000001
1,146,408 364,913 3.14159265359 0
4,272,943 1,360,120 3.14159265359 0
1,725,033 5,419,351 3.14159265359 0
Exploring π
The Internet provides a wealth of information about π. For a list of formulas that use π see note (6). You
may download millions of digits if you are interested and a text editor (Word) may be used to explore the
digits as you wish. Appendix A shows 10,000 digits. If you copy the desired digits into Word you may
use the word counting, find, and the find and replace functions to discover a few of the properties of this
irrational number. Highlight the digits before the ones of interest and use the word count function to
determine the location in the π string.
One question to ask is the distribution of the digits. If the digits 0 through 9 were evenly distributed there
would be 1,000 of each in 10,000 digits. I searched the first 10,000 digits for each digit and replaced it
with the same digit. Word returns the number of replacements made which is the count of each digit.
Table 5 shows the results. This is a trivial observation and Bailey(2) (1988) found that the first 30 million
digits are very uniformly distributed. I wonder how many hours it would take to use Word to make a
similar table for 30 million digits.
Table 5 – Digit Occurances in First 10,000 Digits of π

Digit Occurrences Digit Occurrences
1 1,026 6 1,020
2 1,021 7 970
3 975 8 948
4 1,012 9 1,014
5 1,046 0 968
The highest digit count = 1,046, the lowest = 968, and the mean (of the occurances) is 1000. The standard
deviation, σx, (population) = 30.3743312684.

One of the interesting observations usually mentioned is the occurance of six consecutive 9’s starting at
the decimal position of 762(3) (1-1/2 inches from the left on line six of Appendix A). Consecuctive digits
of five or more do not occur for 0 to 9. My four digit house address occurs twice in the first 10,000 digits.
The series of the odd digits 13579 does not appear, but the even digits 24680 appear twice; 17th line and
the 6th line from the bottom. These are highlighted in yellow in Appendix A. For the reader interested in
exploring the digits of π there is a serious dedicted website designed to do this. See Note (4). One
advantage of searching your own word file is finding multiple occurances of a string.
Are the digits of π random(7)? Speculation is that they are. Should the digits of π be used as an encoding
key? Probably not because they are so “common.”
How may you memorize long strings that contain the digits of π? Usually this is done using a “poem” or
specifically, a piem. Here is one devised by Sir James Jeans.
How I want a drink, alcoholic of course, after the heavy lectures involving quantum mechanics.
The number of letters in each word is a digit of π. A more refined and very long (thousands of words)
piem starts:
Poe3, E1.
Near4 a1 Raven5
Midnights9 so2 dreary6, tired5 and3 weary5,
Silently8 pondering9 volumes7 extolling9 all3 by2-now3 obsolete8 lore4.
During6 my2 rather6 long4 nap3 – the3 weirdest8 tap3!
An2 ominous7 vibrating9 sound5 disturbing0 my2 chamber's8 antedoor8.
"This4", I1 whispered9 quietly7, "I1 ignore6".
This verse provides the first 42 digits of π. - Mike Keith, First verse of Near a Raven (with respects to
Edgar Allan Poe).
Randomly π
The French scientist Georges Buffon (1707 – 1788) conducted what he called his needle experiment to
calculate π which caused quite a stir in his day. Here is how it is described. If we have a uniform grid of
parallel lines, unit distance apart and if we drop a needle of length k< 1 on the grid, the probability that
the needle falls across a line is 2k/π.
The temptation to repeat this experiment is strong and in 1901 Lazzerine is reported to have found that π
= 355/113 = 3.1415929 after 34,080 tosses. This correct six decimal digit value had been previously
determined by Chinese mathematician Zu Chongzhi (429 – 501). It is obvious that “working backwards”
you could stop the number of tosses to achieve the desired accuracy and the 34,080 number seems to
indicate this.
The connection of randomness to π is strong. Note (8) provides a more modern approach to the fun of
using random numbers to predict π. These methods, however, are very very inefficient.

Observations and Conclusions
π is probably the most well-known mathematical constant. The idea that the ratio of the circumference of
any circle to its diameter is a constant has been known for over 4,000 years. Determining the value of π
however, is an ongoing exercise because it is an irrational and transcendental number that never repeats
and never ends.
Memorizing the value of π is a popular pastime for young students and one of the easiest ways to
remember π is to think of the first three odd numbers, 1, 3 & 5. Write them down twice, 113355 and
divide the last three by the first three to get 3.141592^9205 on a 12 digit calculator. You used six digits to
get the answer and your result is correct to a truncated six decimal digits.
The review provides several resources and links to websites that are especially useful. Note (1) shows a
list of mathematical constants that will or have been discussed in this math review series. The next one
just may be the most interesting number you could put into a calculator display.
An appendix (A) includes a single page of the first 10,000 digits of π with a few of my observations of
“interesting” sequences including the Feynman point, note (3), which occurs at decimal digit 762. Word
(or other word processor) may be used to search through and discover various series of digits from a down
loaded file of digits from the Internet. One website, note (4), is especially designed to search 200 million
digits of π very very quickly. Reading through the techniques and methods used for this project provides
an interesting perspective on the fun aspect of π,
____________________________________________________________________________________
Notes for Mathematical Constants – π
(1). A selected short list of the first ten of the more common 47 mathematical constants from Wikipedia
is shown below. # 1 – 3, & 5 – 8 have been discussed in previous math review articles. #10 will be
discussed in Issue 26.
First # of Known
# Symbol Value Name Field N
Described Digits
1 c. 7th–5th
0 =0 Zero Gen R N/A
century BC
2 1 =1 One, Unity Gen R N/A
3 Gen,
i = Imaginary unit A 16th century N/A
Ana
4 ≈ 3.14159 26535 89793
Pi, Archimedes' constant or Gen, 5,000,000,
π 23846 26433 83279 T by c. 2000 BC
Ludolph's number Ana 000,000[30]
50288
5 ≈ 2.71828 18284 59045 Napier's constant, or Euler's
Gen, 100,000,
e 23536 02874 71352 number, base of Natural T 1618
Ana 000,000
66249 logarithm
6 ≈ 1.41421 35623 73095
Pythagoras' constant, square 137,438,
√2 04880 16887 24209 Gen A by c. 800 BC
root of 2 953,444
69807

First # of Known
# Symbol Value Name Field N
Described Digits
7 ≈ 1.73205 08075 68877
Theodorus' constant, square
√3 29352 74463 41505 Gen A by c. 800 BC
root of 3
87236
8 ≈ 2.23606 79774 99789
√5 69640 91736 68731 square root of 5 Gen A by c. 800 BC 1,000,000
27623
9 ≈ 0.57721 56649 01532
Gen, 14,922,
γ 86060 65120 90082 Euler–Mascheroni constant 1735
NuT 244,771
40243
10 ≈ 1.61803 39887 49894
by 3rd century 100,000,
φ 84820 45868 34365 Golden ratio Gen A
BC 000,000
63811
(2). See a much larger table of the distribution of π digits at: Weisstein, Eric W. "Pi Digits." From MathWorld--A
Wolfram Web Resource. http://mathworld.wolfram.com/PiDigits.html This source discusses many interesting
patterns that occur in the decimal digits of π. A list of references is included.
(3). From the (2) source. “The sequence 9999998 occurs at decimal 762 (which is sometimes called the Feynman
point; Wells 1986, p. 51). This is the largest value of any seven digits in the first million decimals.
(4). The best website to search the digits of π to find your

birth date etc. is: http://www.angio.net/pi/piquery You will
also find a lot of trivia regarding the many observations
made on the trivia of what has been found in the first 200
million digits Topics such as Self-locating Strings, The
Meaning of Life (42), Repeating Patterns, How Many digits
of π are required to find a birthday (60,872), and Loop sequences are described. Above right and below are two
search examples. This site also describes how the packed data base of 200,000,000 digits works.
(5). See http://en.wikipedia.org/wiki/Pi for additional information about π.
(6). A very nice list of formulas involving π may be found at: http://en.wikipedia.org/wiki/List_of_formulae_involving_%CF%80
(7). Here is a 2005 research study announcement that addresses using the digits of π as a source of random
numbers. The ScienceDaily article also provides some insights as to the methods used to determine
randomness. "Our work showed no correlations or patterns in pi's number set – in short, pi is indeed a good
source of randomness," Fischbach said. "However, there were times when pi's performance was outdone by the
RNGs."
"These tests are simple to reproduce with a desktop computer. All you need is time," he said. "It took us almost a
year of work to crunch these numbers. We have included the program we used in the paper if anyone would like to
try doing the analysis with a larger number set. I hope someone will because pi shows up in security systems,

cryptography and other places that have nothing to do with circles. That's part of what gives it a fascination that
will not go away." http://www.sciencedaily.com/releases/2005/04/050427094258.htm
(8). Here is a link for the results of a more modern approach to using random numbers to calculate π. The method
uses Python for the calculations.
http://www.stealthcopter.com/blog/2009/09/python-calculating-pi-using-random-numbers/
About the Author

Richard J. Nelson has written hundreds of articles on the subject of HP’s
calculators. His first article was in the first issue of HP 65 Notes in June 1974.
He became an RPN enthusiast with his first HP Calculator, the HP-35A he
received in the mail from HP on July 31, 1972. He remembered the HP-35A in
a recent article that included previously unpublished information on this calcul-
ator. See http://hhuc.us/2007/Remembering%20The%20HP35A.pdf
He has also had an article published on HP’s website on HP Calculator Firsts.
See http://h20331.www2.hp.com/Hpsub/cache/392617-0-0-225-121.html.
HP Solve # 25 Page73 Page 10 of 12

Appendix A – First 10,000 Digits of π – Page 1 of 1.
3.
14159265358979323846264338327950288419716939937510582097494459230781640628620899862803482534211706798214808651328230664709384460955058223172535940
81284811174502841027019385211055596446229489549303819644288109756659334461284756482337867831652712019091456485669234603486104543266482133936072602
49141273724587006606315588174881520920962829254091715364367892590360011330530548820466521384146951941511609433057270365759591953092186117381932611
79310511854807446237996274956735188575272489122793818301194912983367336244065664308602139494639522473719070217986094370277053921717629317675238467
48184676694051320005681271452635608277857713427577896091736371787214684409012249534301465495853710507922796892589235420199561121290219608640344181
59813629774771309960518707211349999998372978049951059731732816096318595024459455346908302642522308253344685035261931188171010003137838752886587533
20838142061717766914730359825349042875546873115956286388235378759375195778185778053217122680661300192787661119590921642019893809525720106548586327
88659361533818279682303019520353018529689957736225994138912497217752834791315155748572424541506959508295331168617278558890750983817546374649393192
55060400927701671139009848824012858361603563707660104710181942955596198946767837449448255379774726847104047534646208046684259069491293313677028989
15210475216205696602405803815019351125338243003558764024749647326391419927260426992279678235478163600934172164121992458631503028618297455570674983
85054945885869269956909272107975093029553211653449872027559602364806654991198818347977535663698074265425278625518184175746728909777727938000816470
60016145249192173217214772350141441973568548161361157352552133475741849468438523323907394143334547762416862518983569485562099219222184272550254256
88767179049460165346680498862723279178608578438382796797668145410095388378636095068006422512520511739298489608412848862694560424196528502221066118
63067442786220391949450471237137869609563643719172874677646575739624138908658326459958133904780275900994657640789512694683983525957098258226205224
89407726719478268482601476990902640136394437455305068203496252451749399651431429809190659250937221696461515709858387410597885959772975498930161753
92846813826868386894277415599185592524595395943104997252468084598727364469584865383673622262609912460805124388439045124413654976278079771569143599
77001296160894416948685558484063534220722258284886481584560285060168427394522674676788952521385225499546667278239864565961163548862305774564980355
93634568174324112515076069479451096596094025228879710893145669136867228748940560101503308617928680920874760917824938589009714909675985261365549781
89312978482168299894872265880485756401427047755513237964145152374623436454285844479526586782105114135473573952311342716610213596953623144295248493
71871101457654035902799344037420073105785390621983874478084784896833214457138687519435064302184531910484810053706146806749192781911979399520614196
63428754440643745123718192179998391015919561814675142691239748940907186494231961567945208095146550225231603881930142093762137855956638937787083039
06979207734672218256259966150142150306803844773454920260541466592520149744285073251866600213243408819071048633173464965145390579626856100550810665
87969981635747363840525714591028970641401109712062804390397595156771577004203378699360072305587631763594218731251471205329281918261861258673215791
98414848829164470609575270695722091756711672291098169091528017350671274858322287183520935396572512108357915136988209144421006751033467110314126711
13699086585163983150197016515116851714376576183515565088490998985998238734552833163550764791853589322618548963213293308985706420467525907091548141
65498594616371802709819943099244889575712828905923233260972997120844335732654893823911932597463667305836041428138830320382490375898524374417029132
76561809377344403070746921120191302033038019762110110044929321516084244485963766983895228684783123552658213144957685726243344189303968642624341077
32269780280731891544110104468232527162010526522721116603966655730925471105578537634668206531098965269186205647693125705863566201855810072936065987
64861179104533488503461136576867532494416680396265797877185560845529654126654085306143444318586769751456614068007002378776591344017127494704205622
30538994561314071127000407854733269939081454664645880797270826683063432858785698305235808933065757406795457163775254202114955761581400250126228594
13021647155097925923099079654737612551756751357517829666454779174501129961489030463994713296210734043751895735961458901938971311179042978285647503
20319869151402870808599048010941214722131794764777262241425485454033215718530614228813758504306332175182979866223717215916077166925474873898665494
94501146540628433663937900397692656721463853067360965712091807638327166416274888800786925602902284721040317211860820419000422966171196377921337575
11495950156604963186294726547364252308177036751590673502350728354056704038674351362222477158915049530984448933309634087807693259939780541934144737
74418426312986080998886874132604721569516239658645730216315981931951673538129741677294786724229246543668009806769282382806899640048243540370141631
49658979409243237896907069779422362508221688957383798623001593776471651228935786015881617557829735233446042815126272037343146531977774160319906655
41876397929334419521541341899485444734567383162499341913181480927777103863877343177207545654532207770921201905166096280490926360197598828161332316
66365286193266863360627356763035447762803504507772355471058595487027908143562401451718062464362679456127531813407833033625423278394497538243720583
53114771199260638133467768796959703098339130771098704085913374641442822772634659470474587847787201927715280731767907707157213444730605700733492436
93113835049316312840425121925651798069411352801314701304781643788518529092854520116583934196562134914341595625865865570552690496520985803385072242
64829397285847831630577775606888764462482468579260395352773480304802900587607582510474709164396136267604492562742042083208566119062545433721315359
58450687724602901618766795240616342522577195429162991930645537799140373404328752628889639958794757291746426357455254079091451357111369410911939325
19107602082520261879853188770584297259167781314969900901921169717372784768472686084900337702424291651300500516832336435038951702989392233451722013
81280696501178440874519601212285993716231301711444846409038906449544400619869075485160263275052983491874078668088183385102283345085048608250393021
33219715518430635455007668282949304137765527939751754613953984683393638304746119966538581538420568533862186725233402830871123282789212507712629463
22956398989893582116745627010218356462201349671518819097303811980049734072396103685406643193950979019069963955245300545058068550195673022921913933
91856803449039820595510022635353619204199474553859381023439554495977837790237421617271117236434354394782218185286240851400666044332588856986705431
54706965747458550332323342107301545940516553790686627333799585115625784322988273723198987571415957811196358330059408730681216028764962867446047746
49159950549737425626901049037781986835938146574126804925648798556145372347867330390468838343634655379498641927056387293174872332083760112302991136
79386270894387993620162951541337142489283072201269014754668476535761647737946752004907571555278196536213239264061601363581559074220202031872776052
77219005561484255518792530343513984425322341576233610642506390497500865627109535919465897514131034822769306247435363256916078154781811528436679570
61108615331504452127473924544945423682886061340841486377670096120715124914043027253860764823634143346235189757664521641376796903149501910857598442
39198629164219399490723623464684411739403265918404437805133389452574239950829659122850855582157250310712570126683024029295252201187267675622041542
05161841634847565169998116141010029960783869092916030288400269104140792886215078424516709087000699282120660418371806535567252532567532861291042487
76182582976515795984703562226293486003415872298053498965022629174878820273420922224533985626476691490556284250391275771028402799806636582548892648
80254566101729670266407655904290994568150652653053718294127033693137851786090407086671149655834343476933857817113864558736781230145876871266034891
39095620099393610310291616152881384379099042317473363948045759314931405297634757481193567091101377517210080315590248530906692037671922033229094334
67685142214477379393751703443661991040337511173547191855046449026365512816228824462575916333039107225383742182140883508657391771509682887478265699
59957449066175834413752239709683408005355984917541738188399944697486762655165827658483588453142775687900290951702835297163445621296404352311760066
51012412006597558512761785838292041974844236080071930457618932349229279650198751872127267507981255470958904556357921221033346697499235630254947802
49011419521238281530911407907386025152274299581807247162591668545133312394804947079119153267343028244186041426363954800044800267049624820179289647
66975831832713142517029692348896276684403232609275249603579964692565049368183609003238092934595889706953653494060340216654437558900456328822505452
55640564482465151875471196218443965825337543885690941130315095261793780029741207665147939425902989695946995565761218656196733786236256125216320862
86922210327488921865436480229678070576561514463204692790682120738837781423356282360896320806822246801224826117718589638140918390367367222088832151
37556003727983940041529700287830766709444745601345564172543709069793961225714298946715435784687886144458123145935719849225284716050492212424701412
14780573455105008019086996033027634787081081754501193071412233908663938339529425786905076431006383519834389341596131854347546495569781038293097164
65143840700707360411237359984345225161050702705623526601276484830840761183013052793205427462865403603674532865105706587488225698157936789766974220
57505968344086973502014102067235850200724522563265134105592401902742162484391403599895353945909440704691209140938700126456001623742880210927645793
10657922955249887275846101264836999892256959688159205600101655256375678

Appendix B – References – Page 1 of 1.
1. The most famous book about π was written by Dr. Petr Beckmann, (November 13, 1924, Prague,
Czechoslovakia - August 3, 1993, Boulder, Colorado) a research scientist, statistician, and physicist
who has authored 14 books and more than 60 scientific papers, mostly on probability theory. A
prolific author he founded Golem Press in 1967 to publish nine of his books. The covers of A History of Pi,
200 pages, ISBN 0-911762-12-4, are shown in Figs. A1 & A2.
RJN photo. RJN photo
Fig. B1 – 1971 hardbound Golem Press Edition. Fig. B2 – 1971 softbound Golem Press Edition.
2. A second good reference, but expensive, is Pi: A source book

(Hardcover) by Lennart Berggren, Peter Borwwdin, and Jonathan
Borwein. The book a source book and is not necessarily a book on
the history of π or a compendium of computation methods. Pi: A
source book is a part of the π lore and it is more acedemic in nature
by its Canadian authors.
Fig. B3 – 2004 hardbound Edition.
HP Solve # 25 Page 75 Page 12 of 12 Last page of issue 25

HP Calculator

Uploaded by

Copyright:

Available Formats

HP Calculator

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HP Calculator

Uploaded by

Copyright:

Available Formats

What are some of the articles discussed in the newsletter?

What are some of the articles discussed in the newsletter?

What anniversary is being celebrated for the HP-12C calculator?

What anniversary is being celebrated for the HP-12C calculator?

HP Solve

Calculating solutions powered by HP

» Calculator » Converting Decimal Learn more about

» Limited Edition HP 15c » How Large is 10^99? » Fundamentals of

» Update Profile » Change Email » HP Home » Support & Drivers

In 1981 HP introduced the 12c---so perfect that

HP Solve # 25 Page 5 Page 2 of 2

 Previous Article – Next 

Introduced to the market in 1982 as a part of the HP

HP Solve # 25 Page 8 Page 2 of 2

 Previous Article – Next 

any other consumer electronics product ever made).

Fig. 2 – The HP-12C joins eight other voyager series calculators.

HP Solve # 25 Page 10 Page 1 of 15

The HP-12C has a long battery life: The HP-12C is famous

HP Solve # 25 Page 11 Page 2 of 15

What has happened in 30 years?

Major changes have been made to the electronics circuits.

The original microprocessor is no longer made and a newer

HP Solve # 25 Page 12 Page 3 of 15

Table 1 – Country of HP-12C Manufacturer with Estimated Dates

1. USA1 SN ’A” September* 1981 to 2003

* - Obviously calculators were made several weeks or months prior

Even business people need to play.

A penny for your thoughts

HP Solve # 25 Page 13 Page 4 of 15

Observations and conclusions

HP Solve # 25 Page 14 Page 5 of 15

HP Solve # 25 Page 15 Page 6 of 15

Appendix A –1982 HP Catalog Page Showing New HP-12C

HP Solve # 25 Page 16 Page 7 of 15

HP Solve # 25 Page 17 Page 8 of 15

HP Solve # 25 Page 18 Page 9 of 15

HP Solve # 25 Page 19 Page 10 of 15

HP Solve # 25 Page 20 Page 11 of 15

HP Solve # 25 Page 21 Page 12 of 15

HP Solve # 25 Page 22 Page 13 of 15

HP Solve # 25 Page 23 Page 14 of 15

HP Solve # 25 Page 24 Page 15 of 15

The scientific programming language of the institutional computer was a higher-level

Programming, what is it?

HP Solve # 25 Page 26 Page 1 of 5

Programming online or offline?

HP Solve # 25 Page 27 Page 2 of 5

Off line program storage

Calculator Programmability Observations and Conclusions

HP Solve # 25 Page 28 Page 3 of 5

Appendix A – HP-15C Programs in the Owner’s &

Owner’s Handbook - 3/82. (00015-90001) http://hp15c.org/hp15c.pdf

HP Solve # 25 Page 29 Page 4 of 5

Advanced Functions Handbook - REV B, 8/84. (00015-90011) http://hp15c.org/hp15cAdvanced.pdf

1. Calculating a more accurate λ = ln(1 + x), pages 181

2. Calculating angle (cosine law) in a triangle, error

3. Accurate roots to real quadratic equation - real and

4. Most accurate variation of #3, uses ∑+ & ∑-. Certain

HP Solve # 25 Page 30 Page 5 of 5

The number of atoms of ANY substance in a volume is: # of atoms = Ndv/