Prefix ERG

News: Registration is free here at the CPMF!

As an informal researcher with access to data in the SNDB. it is part of my mandate to provide information about the notes that our contributors have entered.  Most prefixes contain 10 million notes, with some obvious exceptions like changeover prefixes.  When there is evidence that a prefix is not printed completely, many collectors and other researchers would likely want to know about it.

Prefix ERG is a Journey series $20 prefix printed in 2008.  I have prepared a short summary of the serial number data as follows:

ERG 0000000-1799999 : 227 notes in SNDB (33% of all notes entered)
ERG 1800000-2879999 : 0 notes
ERG 2880000-2913749 : 5 notes in SNDB (<1% of all notes entered)
ERG 2913750-6629999 : 0 notes
ERG 6630000-9999999 : 455 notes in SNDB (66% of all notes entered)

So, as you can see, there are only 5 notes recorded over 1800000 and below 6630000, and those 5 notes form a cluster.  I would consider the data for prefix ERG to be mature at this point in time, meaning that I don't believe that notes will be found in the ranges where no notes have been found to date.

The 5 notes that form the cluster have the following serial numbers and position numbers:

ERG 2887895 10/14
ERG 2889245 17/23
ERG 2895443 35/49
ERG 2895542 35/49
ERG 2911112 11/48

Not coincidently, I found ERG 2889245 17/23 today.

I have determined that these 5 notes were not printed with the usual skip interval of 8,000 that is common for nearly all $20 notes printed by the Canadian Bank Note Company.  The skip interval appears to be 750, and if all 45 positions of the sheet layout were used, the interval ERG 2880000-2913749 represents a range of only 33,750 notes.

Had prefix ERG been printed normally, notes in the range ERG 1800000-2879999 would represent 3 normal reams printed with a skip interval of 8,000.  So at ERG 2880000, another ream would be expected to start and run all the way to ERG 3239999, representing a ream of 360,000 notes.  But of course, we don't see that in the data.  Normal production of notes does not start again until ERG 6630000, and that's not a "normal" starting point for any ream.

So, in conclusion, prefix ERG is not a complete prefix, meaning it does not have 10 million notes printed with it.  It has only just over 5.2 million notes.  This makes ERG a short prefix, but certainly not a rare one (except in the range 1800000-6629999).  As expected, there is no explanation for why we are seeing the odd distribution of serial numbers at the start and the end of the prefix.  Had there been absolutely no notes found in the range ERG 1800000-6629999, then there is the possibility that these notes are sitting in some vault somewhere waiting to be released if they have not been destroyed leading up to the release of the new polymer series.  But because notes in the range ERG 2880000-2913749 have been found, and because we know that they were printed differently (i.e., with a skip interval of only 750), then it becomes extremely unlikely, in the humble opinion of this researcher, that notes in the ranges ERG 1800000-2879999 and ERG 2913750-6629999 ever existed or, had they existed, were ever slated for release.

I have ERG 2889245 in my hand, and it is a moderately circulated note that would grade VF.  It appears to be an ordinary note with no unusual features.  I am going to hold on to it for now, and maybe in time, we will find out if the Bank of Canada released any test notes in prefix ERG. 

Your comments and questions are welcome.

Very Interesting. Thank you for sharing. ERG0 can serve as a word prefix if 0 (zero) were to be taken as an O (oh).

ERG0 can serve as a word prefix if 0 (zero) were to be taken as an O (oh).

Would be a good mate for THU5 (if 5 were to be taken as an S...and assuming that we would get up to the letter "T" in the prefixes for $50 notes). 

Very Interesting. Thank you for sharing. ERG0 can serve as a word prefix if 0 (zero) were to be taken as an O (oh).

ERG is a word.  You crossword lovers out there know what I mean. 

I found your observations interesting mmars (as I'm sure many others do).  I believe we would find similar observations if more data were collected on some of the 2003 $10 prefixes, $10 FEV, or 2004 $5 HOH prefixes (noted in Charlton) and others.   

It it led me to wonder 2 other things:

1) was there any geographical correlation between the entries of ERG prefixes & their distribution (eg: notes less than 1.79 M 20% found in western provinces, 30% found in ON, 30% found in Quebec, 20% eastern provinces, etc)

and
2) is there any way the SNDB could notify a collector that a prefix is being investigated such as the ERG?  One of my favourite features of the SNDB is the way it notes whether a note is an insert, a radar, or repeater,etc but I noticed it will not give a notification if the note is a short prefix, change-over, or other significance.  I know this may be asking a lot but I was wondering if that could be done (for us who don't have access the SNDB data).

but now I got to go watch the Leaf hockey game....

Just thinking about one thing. How does the fact that forum members are generally not encouraged (or even continuous notes being removed from SNDB) to enter continuous notes play into this type of analysis.

I have always felt that not entering continuous notes creates a missed opportunity in terms of finer analysis in such cases. I realize that it creates a lot of notes to manage particularly if the main interest is that of finding the print arrangement.

Incidentally is the $20 polymer note from CBN having different print layout then $100 (Perhaps this should be a separate topic?) I am seeing something weird but then it may just be my ignorance of these things.

I have always felt that not entering continuous notes creates a missed opportunity in terms of finer analysis

Can you imagine if every bricker entered runs of a 1000, 4000, or 16000 notes weekly...

Just thinking about one thing. How does the fact that forum members are generally not encouraged (or even continuous notes being removed from SNDB) to enter continuous notes play into this type of analysis.

I have always felt that not entering continuous notes creates a missed opportunity in terms of finer analysis in such cases. I realize that it creates a lot of notes to manage particularly if the main interest is that of finding the print arrangement.

I can't speak for the creator and head administrator of the SNDB, but my guess is there is simply no easy way to get brick data into the database.  When the SNDB was new, we tried reasoning with the brick people about creating a new way to input brick data, but the co-operation was not there.  Brick people only trusted Gilles Pomerleau with their finds.  And I don't want to step on any toes, but people with access to large numbers of brand new notes don't like letting everyone know they have access to large numbers of brand new notes, and this is true for several reasons, not just the obvious reason of personal security.

Runs of consecutive notes always have the same position numbers.  At one time, bricks were made up of 1,000 sequentially-numbered notes, and because both printing companies used skip intervals of 1,000 or more, every note in a brick had the same position numbers.  Nowadays, with mixed-up bricks becoming the norm, bricks contain  many smaller runs of notes covering a range of serial numbers that can be in the hundreds of thousands.  But the sequential runs will still have all the same position numbers.

Statistical analysis requires random sampling, and that's why circulated notes are preferred.  Circulated notes are pretty close to random, though there are going to be regional disparities as well as a non-uniform distribution of people reporting notes.  There is absolutely nothing random about bricks, even mixed-up ones.  This is why we really don't want to see more than the end notes of every sequential run that people find.

Here's some mathematics to help explain things.

Most prefixes contain 10,000,000 notes.  At 1,000 notes per brick, that's 10,000 bricks per prefix.  For any given prefix, the SNDB receives a couple of thousand entries.  So we get roughly 0.2 notes per brick.  For our research purposes, we don't even need that level of detail.  That's why we're really not crazy about getting brick data into the database.  Even in ranges where standard numbering practices used by the printers deviate from normal, the brick data is not really helpful.

Edited passage in bold

1) was there any geographical correlation between the entries of ERG prefixes & their distribution (eg: notes less than 1.79 M 20% found in western provinces, 30% found in ON, 30% found in Quebec, 20% eastern provinces, etc)

I don't have access to geographical information collected by the SNDB.  But in theory, there are going to be geographical biases for any and all prefixes.  That's just logical.

and
2) is there any way the SNDB could notify a collector that a prefix is being investigated such as the ERG?  One of my favourite features of the SNDB is the way it notes whether a note is an insert, a radar, or repeater,etc but I noticed it will not give a notification if the note is a short prefix, change-over, or other significance.  I know this may be asking a lot but I was wondering if that could be done (for us who don't have access the SNDB data).

It sounds like what you're asking for is that the SNDB have perfect hindsight.  The data collected is used to find irregularities.  We don't know ahead of time if a prefix is going to have irregularities.  We would also shy away from speculating on irregularities until there is enough evidence to put forward observations and draw temporary conclusions.  I don't know how useful it would be to the individual entering notes into the SNDB to receive a message saying that a certain prefix is somehow different.  We don't investigate prefixes in a deliberate manner.  We usually find irregularities after combing through large amounts of data, and collecting data takes time, meaning that the data needs to be mature enough to form the basis for doing specialized research.  In other words, a prefix has to be on its way out of circulation before the data for that prefix becomes meaningful.  I'm sorry, but that's just a logical approach to things.  We don't want to be like some of the hucksters on Internet auctions proclaiming the rarity of items that have just been released.  For example, $20 polymer notes could be said to have been extremely rare on their first day of issue to the public, and that's why people who got the first ones from their banks tried selling their notes at a premium.

Suggestions for how to improve the SNDB in the future can always be addressed to the SNDB owner.  For example, if you want to see a message when an ERG note is entered, I'm sure it's a possibility, but that's not my decision to make.

But the sequential runs will still have all the same position numbers.

Thank you so much for your detailed explanation. It helps a lot. But here is where I started to run into difficulties.

I have run into a following sequence with CBN polymer notes (not yet entered in the SNDB):

I have run into the following serial numbers are for the same prefix -

xxx2233 - Plate number 01
xxx2770 - Plate number 03

In the above range I had several notes in sequential order.

So these notes are  within the same thousandth i.e. 2000 to 2999 but with different plate numbers. If I follow the guideline of SNDB to only enter one note from the sequence of notes within a brick pattern then should I enter xxx2233 Plate number 01 or xxx2770 plate number 03?

Say for arguments sake I entered both how does the SNDB know that I have not made an error in data entry?

Notwithstanding the fact that consecutive notes are only to be entered once the above finding raises at least two further questions:

a) Is it possible that CBN is using skip interval of 500?
b) Is it possible that if skip intervals of 1000 or more are used then the numbering of notes printed in a run begins and ends with something other than 000 and 999?

The bigger question is this:

How does the SNDB in case of polymer $50 and polymer $20 deal with data entry errors? For other notes if I entered a wrong plate number it tells me that the Serial number and Plate number combinations are not valid.

I do not mean to hijack the ERG topic so if there is a way to move this to another topic let me know but these items do seem related.

Thank you so much for your detailed explanation. It helps a lot. But here is where I started to run into difficulties.

I have run into a following sequence with CBN polymer notes (not yet entered in the SNDB):

I have run into the following serial numbers are for the same prefix -

xxx2233 - Plate number 01
xxx2770 - Plate number 03

In the above range I had several notes in sequential order.

So these notes are  within the same thousandth i.e. 2000 to 2999 but with different plate numbers. If I follow the guideline of SNDB to only enter one note from the sequence of notes within a brick pattern then should I enter xxx2233 Plate number 01 or xxx2770 plate number 03?

You should enter both.  I think the guidelines state you can enter the first and last notes in every sequential run, even when those runs occur in the same "thousandth".

Really, the guidelines for the SNDB exist as a substitute for common sense.  People with common sense don't need to be told what they should and shouldn't do.  But, unfortunately, common sense is not that common any more, and we have to deal with all kinds of people who would enter faulty or redundant data for whatever number of reasons. If the guidelines did not exist, then it would just make dealing with trouble-makers that much more difficult.  Reasonable people know how to conduct themselves.

Say for arguments sake I entered both how does the SNDB know that I have not made an error in data entry?

It doesn't.  All it can do is reject position number pairs that don't exist.

Notwithstanding the fact that consecutive notes are only to be entered once the above finding raises at least two further questions:

a) Is it possible that CBN is using skip interval of 500?

Yes.  That is possible.

 

b) Is it possible that if skip intervals of 1000 or more are used then the numbering of notes printed in a run begins and ends with something other than 000 and 999?

Reams don't have to start or end on those numbers, but 99.99% of the time, they do.

The bigger question is this:

How does the SNDB in case of polymer $50 and polymer $20 deal with data entry errors? For other notes if I entered a wrong plate number it tells me that the Serial number and Plate number combinations are not valid.

If a position number pair is not valid, the SNDB rejects the entry.  Sometimes, though, a position number pair will be valid but wrong because of a mistake being made by the person entering the data.  One of the more common mistakes is people reverse the face- and back-position numbers since these combinations occur as pairwise reciprocals on Journey series notes. (For example, a person will enter 78/58 instead of 58/78, though both combinations are valid and exist.)  Errors make it into the database, and that's fine because nobody is perfect.  Errors are expected, and only where they are numerous do we take a closer look to see if mischief is going on.

Now, I have answered your questions, and I have some for you...

Can you be more specific about the notes you found? i.e.,
xxx2233 - Plate number 01
xxx2770 - Plate number 03
Can you reveal, either on the forum or via PM, the prefix and first three serial numbers of your notes?  I'm just curious and have a hunch about your notes.

it sounds like what you're asking for is that the SNDB have perfect hindsight.

That would only be the case if I were asking for that type of info on the New Frontier Series (since the data is unknown).    I suppose asking for data on investigated prefixes is asking too much (& I do not wish to encourage internet "huskers" speculating any more than they are) but I don't think asking for an indicator that a prefix is a change-over (once it has been established) is asking too much. 

Can you be more specific about the notes you found? i.e.,

Thank you  - Will PM you

Now, I have answered your questions

Thank you for answering the questions. Clearly some of the questions were rhetorical and not meant to suggest that someone would knowingly enter erroneous information. The fact that many forum members spend time and energy having the SNDB set-up and doing research is commendable so this dialog hopefully is  taken in the spirit of educational discussion. 

Data entry errors even in controlled environments of research are real things let alone in a situation like SNDB where information is entered by many. From a purely research methodological perspective I still do not think that answers about errors and dealing with errors were convincing. Perhaps for the expected results from the research they may be inconsequential.

However one question still remains unanswered:

When the plate numbers corresponding to the serial numbers are not know i.e. when skip interval and/or the layout (45/on or 40/on etc.) are not known such as the case is with $20 polymer and $50 polymer for CBN then how does the SNDB recognize data entry errors other than when the FP and BP are not the same or the serial number is not seven digits etc.

However one question still remains unanswered:

When the plate numbers corresponding to the serial numbers are not know i.e. when skip interval and/or the layout (45/on or 40/on etc.) are not known such as the case is with $20 polymer and $50 polymer for CBN then how does the SNDB recognize data entry errors other than when the FP and BP are not the same or the serial number is not seven digits etc.

The specific methods that the SNDB uses to automatically or manually verify entries will not be disclosed for the reason of preventing fabricated entries.

Found one today. Will enter it in the database tomorrow.