I recently got an old 42s from a colleague and am having blast playing around with it but now it cannot seem to do any statistical functions.

All Statistics functions as well as trying to clear the statistics registers gives me an "Invalid Type" response.

Unfortunately, I did not get a manual with this. Could I have inadvertently erased the statistics registers themselves?

Thanks

You're most likely in complex mode. Try shift-STO to toggle out of complex.

Try MODES SIZE 25 and see if that fixes it. You should now have 25 storage registers (R00 thru R24). Registers R11 thru R23 are used for statistics summations, unless they have been relocated by the "summation registers" command. I'd strongly recommend that you buy the MoHPC CD with the HP 42s manual on it (CD #1).

Thanks,

It was a combination of SIZE and the variable (Sum)Reg being messed up. I used to have a 41CX so a lot is similar but that was like 15 years ago so I can't remember it all.

Norman --

It seems that you've got it figured out, but here's some background for everyone:

For those HP's that do statistics using the summation method (instead of maintaining all input data, such as the RPL-based models, HP-17B/BII, HP-27S, and HP-71B), a set of six numbered storage registers was utilized for summation.

Most legacy models use a fixed set of registers. The HP-11C and HP-34C used R0-R5, as those would be the last to be deallocated for programming storage. The HP-15C uses R2-R7, because R0 and R1 must be preserved for matrix indices.

The HP-41 utilizes a set of six consecutive registers, user-selectable by the (Sigma)REG function. The HP-41CX added (Sigma)REG? -- which is not available in the X-Functions module -- to allow the user to identify the first summation register. The user is responsible for allocating a sufficient number of registers for numerical storage (SIZE), and that the selected value of (Sigma)REG allows for six summation registers.

The HP-42S, for compatibility with the HP-41, did the same. Both (Sigma)REG and (Sigma)REG? are standard in the HP-42S.

The HP-32S and its descendants do this better than all the others:

They use the pool of unallocated RAM for the statistical registers. Each summation value can be recalled individually, but values are set and updated only via CLEAR(Sigma), (Sigma)+, and (Sigma)-.

-- KS

*Edited: 6 Dec 2009, 5:31 p.m. after one or more responses were posted*

Quote:

The HP-32S and its descendants do this better than all the others:

I agree, completely!

Quote:

Each summation value can be recalled individually, but values are set and updated only via CLEAR(Sigma), (Sigma)+, and (Sigma)-

Not quite, on the the 32sii they are writable via indirect data access. Registers A-Z are indirectly accessible as 1 thru 26, 27 is the "i" register itself and 28 thru 33 are the statistics registers. This turns out to be extremely useful in some programs.

I believe that this feature was new on the 32sii and was not on the 32S, but it continued with the 33s and 35s. The 35s skips the "i" register because it has the "I" and "J" indirect registers. It also requires negative numbers to indirectly address the fixed location registers (A thru Z corresponds to -1 thru -26) and the statistics registers correspond to indirect -27 thru -32.

-Katie

*Edited: 6 Dec 2009, 5:31 p.m. after one or more responses were posted*

Quote:

Not quite, they are writable via indirect data access. Registers A-Z are indirectly accessible as 1 thru 26, 27 is the "i" register itself and 28 thru 33 are the statistics registers. This turns out to be extremely useful in some programs.

*Mea culpa* -- forgot about the indirect addressing.

Yes, having six easily-clearable "scratch" registers available for intermediate results without using lettered registers is helpful, but the required indirect-access storage could be a nuisance.

Quote:

I believe that this feature was new on the 32sii and was not on the 32S, but it continued with the 33s and 35s.

You are right! I didn't know that one; add it to this list of 32S vs. 32SII differences I posted several years ago.

-- KS

*Edited: 7 Dec 2009, 12:19 a.m. after one or more responses were posted*

Quote:

but the required indirect-access storage could be a nuisance

It could be but sometimes it's a big help when you use Sigma+ and Sigma- along with them. Take a look at how I use these registers here.. This program has nothing to do with statistics but makes use of the the extra memory to store processing variables and leaves the registers (1) - (26) available for results.

-Katie

*Edited: 6 Dec 2009, 5:38 p.m. *

It sounds like REGS was made complex. The numbered storage registers on the HP-42S are actually elements of a matrix variable named "REGS", and that matrix can be either real or complex. When it is complex, the statistics functions won't work, giving you the "Invalid Type" message, but you can fix that quickly by doing CLV "REGS", followed by SIZE 0025 to re-create REGS as a real matrix.

If REGS is a real matrix, but one of the summation registers contains an ALPHA string, trying to use the statistics functions will give the "Alpha Data Is Invalid" message, but CLSigma will fix that problem.

- Thomas

*Edited: 7 Dec 2009, 12:05 a.m. *

Quote:

Yes, having six easily-clearable "scratch" registers available for intermediate results without using lettered registers is helpful, but the required indirect-access storage could be a nuisance.

It would have been a better design if the statistics registers had been 27 through 32 and the i register had been 33.

Palmer

Quote:

It would have been a better design if the statistics registers had been 27 through 32 and the i register had been 33.

I agree. HP put a lot of thought, planning and engineering into the 32sii (complex numbers excluded) and this seems to have been a bad decision. However it may have been a deliberate attempt to semi-isolate the statistics registers so that users wouldn't overwrite them so easily when programming.

-Katie

Quote:

It would have been a better design if the statistics registers had been 27 through 32 and the i register had been 33.

Maybe so, but the more fundamental shortcoming is the insufficient RAM. 2 kB -- the amount provided on the original HP-28C -- was too stingy on that model with matrix functions and symbolic algebra, but would have been ideal on the HP-32S/SII. The 390 bytes on the HP-32S and 384 bytes on the HP-32SII do not allow the user to store a library of useful programs and equations, even though the paradigm is well-suited for that purpose. It's as though H-P deliberately "de-contented" the HP-32S/SII to prevent it from siphoning additional sales from the HP-42S.

It makes little sense that storage space for a variable is "deallocated" by setting its value to zero. This means that functions such as SOLVE and INTEG can cease to work for no apparent reason simply because the user had stored a non-zero value in a register. This can't happen on an HP-42S or a pre-Saturn model, where, if space for a register is allocated, it can be used; if it isn't, it can't.

With 2 kB of RAM, the 26 letter variables and the indirect register should have been allocated permanent space of 27 * 8 = 216 bytes. The remaining 2048 - 216 = 1832 bytes for statistical summation (48 bytes), programs, and equations would have been more than enough for most users.

This isn't an issue on the HP-33s and HP-35s, now that RAM is so cheap that quantities beyond what is practical can be provided. Unfortunately, those newer models don't measure up to the HP-32SII in many ways.

-- KS

*Edited: 10 Dec 2009, 1:31 a.m. *