The following errata sheet describes two corrections to the owner's manual for early manufactured HP-67 calculators. The same information is applicable to early manufactured HP-97 calculators. Later-model HP-97 calculators had these errata corrected in their firmware ROMs. It is likely that later-model HP-67 calculators had this correction as well. The errata has been verified on a HP-67 with serial number prefix 1706A, and on a HP-97 with serial number prefix 1704A. The errata has been verified to be fixed on a HP-97S with serial number prefix 1830A.

There are two parts of the errata. The first part has to do with the operation of the LAST X register for inverse trig. functions. The easiest way to test a machine to see if it has the errata or not is to enter -1, [sin-1], LAST X. A HP-67 or HP-97 with the errata will display 1 instead of -1. A later machine with the errata fixed will display -1.

The second part of the errata has to do with an excessive amount of error for certain arguments of inverse sine and cosine functions. There are six specific arguments that produce this error. The errata sheet gives the errors in percentages, following are the six values, the answer on machines with the errata, and the correct answer. These answers are all for the sin-1 function in degrees mode:

X=0.000003000, errata = 1.730332541E-4, correct = 1.718873385E-4

X=0.000004000, errata = 2.349126960E-4, correct = 2.291831181E-4

X=0.000005000, errata = 2.979380535E-4, correct = 2.864788976E-4

X=0.000006000, errata = 3.666939889E-4, correct = 3.437746771E-4

X=0.000007000, errata = 4.343020087E-4, correct = 4.010704566E-4

X=0.000008000, errata = 5.110783533E-4, correct = 4.583662361E-4

The LAST X register is primarily intended for error recovery in the event an undesired function is mistakenly executed; i.e., the "opposite" function can be easily executed on the same number, thus conveniently getting the user back to the previous step in the problem. Furthermore, with trigonometric functions, the inverse operation (sin, sin-1) can be used to directly return to the previous step in the problem. Thus LAST X is not essential for easy error recovery when dealing with trigonometric functions.

**Page 67, LAST X.** Add the following statement
to clarify the description of the LAST X register:

*A table of operations that save X in the LAST X
register is shown in appendix D.*

**Page 323, LAST X.**Add the following note
next to [sin-1], [cos-1], and [tan-1]: * Except
for arguments equal to 0 or -1.*

**Page 92, Trigonometric Functions.** Add the
following paragraphs:

There exist several specific argument values for which sin-1 (and to a lesser degree, cos-1) are in error to an extent that could be excessive for some applications. However, these arguments are very small in magnitude and thus infrequently encountered by most users.

The six specific arguments affected and the resulting errors for sin-1 X are X=0.000003000 (0.6%), 0.000004000 (2.5%), 0.000005000 (4.0%), 0.000006000 (7.0%), 0.000007000 (8.0%), 0.000008000 (11.5%). No other values are affected. Notice that changing the magnitude of the above arguments by as little as +/- 0.000000001 eliminates the larger-than-normal error.

**HEWLETT [HP] PACKARD**

00067-90034 Printed in U.S.A.