Final proofing documentation for TTB

[Distiller Mike]

Hi all, I am trying to find a form of sorts to document proofing for any would be audits and my own documentation. Unfortunately, I can't find one and am resorting to building my own. Does anyone have a general form and or ideas of documentation to track such as, methods used, how many samples taken, temp, etc..? Thanks all!

[dhdunbar]

The regulations (19.618 for other than packages and 19.619 for packages) require a gauge record for any of the required gauges specified.  You can see the information that record must contain.  Did you know, for example, that it is suppose to contain the proof of distillation?  Well, it is. 

However, the gauge record  section does not answer how you record, for posterity, the elements of the gauge.  Obviously, the elements  of the gauge are different if you determine the wine gallons by weight or by volume. By volume, you have two corrections:  one for the built in error of the certified thermometer - for which you have a correction table, of course - and one to correct to the volume at 60 degrees H.  For hydrometer readings, assuming you do not have one of the expensive density meters that TTB accepts, you need the correction for the variance from apparent proof that is intrinsic to the instrument, and for the correction to 60 degrees F.  

If you know how to use the tables in the gauging manual, it is easy to prepare a temple in Excel with formulas that allow you to enter the apparent readings and table values and convert the apparent readings to actual readings.  Some people sell apps for doing that, but I've not checked them out.  

TTB provides a handy  proofing worksheet with the package it sends when it approves an application to establish a DSP.  It doesn't come with instructions, but it shows what they would like to see:

 

Append one of those to each serially numbered gauge record, or to other records, hen no gauge record is required under 19.618 or 19.619, and you will have done everything TTB might require in its wildest dreams of compliance.

[dhdunbar]

Whoa - I misread this the first time.   I bet you mean the proof and fill check you make  at the time of bottling.  I'll talk about that here.  I wish I could do it in a few short words.  I can't.

The law says TTB must make regulations that make sure consumers are not misled as to the quantity of the product in the bottle.  For spirits, the quantity is the volume, but it is also the proof.  So TTB requires that you tell consumer how much and how strong.  Because TTB is not making tests to determine if you are doing that, they require that you keep records to prove that you are.  It sounds good in theory.  It sounds professional.  But it is not. Your records will not provide assurance of compliance.  Let’s look at why I say that.

The regulations provide for a record of the proof and the fill of bottles.   You find the record requirement in 19.600, which is headed "Alcohol content and fill test record."  It requires specific information, but you will not find a "form" that TTB offers as appropriate for your use.  In the broad sense, the records you keep of the tests you perform for each bottling are supposed to allow a TTB official to determine that you are complying with Sec. 19.356.  

Sec. 19.356 requires that, during bottling, you take adequate samples of bottled spirits, at representative intervals, to "ensure" that products are within the tolerances established by that section.  I add the emphasis to point out TTB doesn't tell you how to do what they require.  The regulations do not discuss how many samples you must take and at what intervals you must take them in order to "ensure" compliance.   They place the onus on you. 

Why doesn’t TTB give sample sizes and frequency? It doesn’t do that because it is impossible for you to do what they ask. Sampling is not an easy subject and I am not an expert.  I know just enough to recognize hogwash.  It is hogwash to expect any small distiller to have a sampling program that “ensures” compliance.  It also is hogwash to expect any small distiller to have a sampling program that will allow statistically valid projections of compliance at any reasonable confidence level.

If pressed about sample parameters, the TTB officer who is auditing your records would probably dance to the side with a comment, every case is different, so we can’t specify sample size.  Well, sure, every case is different, but the rules for statistical sampling, i.e., for sampling from which the TTB officer can draw the conclusion required by 19.356, that the sampling program ensures compliance, don’t change.  And that leads to problems. 

The basic rules of statistical sampling do not allow for a sampling program that will ensure compliance.  Talk of ensuring is prattle.  Since TTB cannot possibly create a sample standard that ensures, it punts.  TTB says it is up to you to devise the program. 

Going a little deeper.  If we want to draw a conclusion from a sample, we must know the sample size needed to establish, within an acceptable degree of accuracy, that the bottles are consistently within the tolerance established by 19.356.  Assuming you select the sample in a valid way, given the sample size, the frequency of the sample, the expected error rate, and the universe from which it is obtained, you will have a level of assurance, say 95%, that between 96% and 98% are within the tolerances allowed.   Note that it is a qualitative finding.  It says nothing about how much things vary.  An error of 5% counts just as much as an error of 50%.  All errors are created equal in frequency sampling.

Which is fine, as far as theory goes.  But when the universe is small, i.e., the size of a bottling run by a small distiller, no sample size will provide results that will allow a conclusion of the sort, we are 95% sure that the between 96% and 98% of the bottles were within tolerance.  

Even if TTB were to have a standard that said, you must take sufficient samples to establish, with a 95% confidence level, that between 96% and 98% of the bottles were in tolerance, TTB would have to determine how taking stratified samples, instead of random samples, affect the assurance of compliance, what rate of error is expected, etc., i.e. how to deal with all the variables that bedevil students in college statistics classes, let alone a small distiller left adrift by broadly stated  requirements for which TTB provides no further guidance.

Worse yet, in a small business, audit standards provide that an auditor looking at compliance must assume that the internal controls are inadequate to ensure.  And proof and fill tests are "internal controls."  Because the auditor cannot place any reliance on internal controls (you could be inventing all the test data), the auditor would have to test every bottle to ensure compliance.  But most of those bottles had best be gone from bond or your business is probably in deep trouble.

My point?  Sampling is not an easy subject and I am not an expert.  I know just enough to recognize hogwash.  It is hogwash to expect any small distiller to have a sampling program that ensures compliance.  So, what would I recommend, recognizing that I am not TTB, but that TTB has chosen to punt? The practical answer is, provide the TTB officer with information that satisfies the TTB officer, who, like you, is without an objective standard for how many and how often to test.   I know, that is a statement of the obvious, but it is all we are left with.  My best advice, in practice, is ask yourself, based on your sampling, if you are reasonably sure that you are within tolerance.  If you are reasonably sure and the TTB officer is a reasonable person, your program will pass the test.  That is the best guide I can offer.

I'd be interested in hearing comments from any statisticians and/or auditors on my comments.

 

[Distiller Mike]

Hi and thanks for this @dhdunbar! This is great perspective and insight. Thank you.

 

[Still_Holler]

On 11/16/2017 at 5:57 PM, dhdunbar said:

TTB provides a handy  proofing worksheet with the package it sends when it approves an application to establish a DSP.  It doesn't come with instructions, but it shows what they would like to see:

Do you know of any form/worksheet like this for Gauge Record? I thought I read on one of these forums there where a downloadable set of templates that came with becoming an ADI member but now cant find it.

[dhdunbar]

Not a member, so I don't know.  TTB does not supply them.  You've got two gauge records to deal with - bulk gauge and package gauge.  

[dhdunbar]

I'll add something here.  You need to make sure that the gauge record records the kind of barrel into which you place the spirits.

If you examine what is required in the storage record, you will find that you must keep tank records and package records.  If you examine the information that the package record must include, you will find, much to my surprise, I'll add, that it does not include the type of barrel in which the spirits were stored.  In general, you must keep records of the receipt into and activities in the storage account (19.590).  Focusing on barrels, you record spirits received into the account in barrels, filled into barrels from a tank, and transferred from one barrel to another, and the addition of oak chips, if any. 

You must keep package summary records (19.591).  A package summary record is a separate records of each kind of spirit – separated too by domestic and imported and VI or Puerto Rican origins - in packages that show the spirits deposited in, withdrawn from, and in the storage account.  The regulations then require that they be arranged in a certain order.  For domestic spirits, that is alphabetically by State and by the plant number and name of the producer or warehouseman.  For most of you, that is easy probably easily enough done with a sortable spreadsheet or database, or not a problem because you only  get spirits from one or two sources.  But I mention it to show what is required, but also what is not.

The package summary records must show the date of the transaction, the number of packages and proof gallons covered by the summary record,and any gains or shortages disclosed by inventory or when an account is closed (19.591) and the balance on the summary records (remember this is for each type of spirit) in the account at the end of the month.  These are the figures that feed into the report of storage operations you send to TTB each month.

As an aside, you are not required to inventory spirits that are in packages, so the shortage on inventory would be the loss of a package.  You will record losses when you empty the packages, something the regulations seem to omit.

You must also” consolidate” the records at the end of each month, “, to show, for all types of containers (barrels, totes, tanks, etc.) and kinds of spirits, the total proof gallons received in, withdrawn from and remaining in the storage account.  

That’s it.  Well, you may say, it seems that is certainly more than enough.  But where is the requirement that you show whether the barrel in which you have stored spirits produced from a corn mash are being stored in new charred, used, new uncharred, or used oak, or stainless steel or plastic totes?   Oops, follow these requirements to the letter, and you cannot prove, from these records alone, to the inquisitive TTB,  that the product you are labeling bourbon was stored in new charred oak, or that the product you are labeling corn whiskey was not.  In fact, you can’t even show that it touched oak, so you can’t even prove its whiskey. 

That is where the gauge record comes in, because the gauge record, and as far as I have found, only the gauge record requires that you show the type of package into which you place the spirits.

You must prepare a serial numbered (the serial number is important for tracking purposes) package gauge record (19.619) when you enter spirits for deposit into the storage account, package spirits from a tank in the storage account, The gauge record must include, for each package, among other things, “Cooperage identification (``C'' for charred, ``REC'' for recharred, ``P'' for plain, ``PAR'' for paraffined, ``G'' for glued, or ``R'' for reused, and ``PS'' if a barrel has been steamed or water soaked before filling).  And, because you also must show “the identity of the related transaction form or record, and its serial number – if it has one – your records show that the the bouyrbon was stored in new charred oak. There you have the only required record, again, as far as I can determine, that is going to show that.  Further, because the gauge record is going to show the proof of the spirits, it will also show that the bourbon was stored, as required by the standard of identity (5.22) at 125 prof or less.  I’ll add another aside, when you gauge packages you must do so by weight (19.619).

Finally, although it cannot be used to trace spirits through the system, if you transfers spirits from one package to another, you must give the package the same package identificaitohn number as the original, but must also affix, to the head of the new package,  a sign or label  “in the following form:     The spirits in this ------------ [kind of cooperage: barrel or drum], package identification No. --------, were transferred from a ---- -------- [kind of cooperage: barrel or drum], on -------------- [Date],  -------------- [Proprietor].” (19.468).  Strangely, as far as I can determine, when you first put the spirits into a package, this information is not required.  You need only assign a package identification number (19.485), which does not include any indication of the type of cooperage.  I’ve not found an explanation for this. 

Out of curiosity, I'd be interested in knowing if this sort of compendium, of the various sections, into one place, is of interest to those of you who operate distilled spirits plants? Would it be useful, for example, to have, in one place, a description of all of the information you need to know if you are bottling bourbon, making bitters, using wine as a flavoring agent, etc?  Does  it answer a need?  And would you pay for that, say, in the form of an online tutorial, from which you could pick, like items off a menu,. as need?  Would, for example, a discussion of vodka be worth $30, one of bourbon $40, and TTB's stance on GRAS,,  something else?  Would an all access option be attractive?    

It would be a chore to do that.  It would take a lot of time.  And like textbooks, it would have a specialized audience - even narrower than a text book - and so could not be published as a book for $50.00.  The ROI on my time would not be there.  This answer, for example, took me a couple of hours to compile in useable, I hope, form.  I did it because I think it is important information.  But I've got a life, both commercial and personal.  If you have an opinion on this, I'd like to hear it by PM, to keep the forum free of liter.  Thanks.