Bob Pisani Revisited (a fresh look)

I recently completed a batch of data importing for the database project (link), which cleared a backlog of files waiting to be processed. The completion is a huge milestone because it finally provides the ability to search (almost instantly) for gold bars across 18 different funds, in records covering the last 18 months. So while Screwtape Files was the FIRST blog to identify Bob Pisani's mystery bar ZJ6752, we now return to the study armed with more information.

The goal remains the same ... "use the database to identify gold bars that we see in the media, for additional narrative". This gives us a larger picture for virtually no extra effort, because the hard work of building the index is already done.

Bob Pisani and me,
from the filmed-but-
never-aired documentary:
"Prosimians in the Attic".
First, a quick history for those who haven't encountered the ZJ6752 story - CNBC's Bob Pisani presented a small series on gold in 2011. We're interested in serial numbers for bars that appear on camera. I studied the clips to find more serial numbers than just ZJ6752. These are the video references (in order most interesting to least), with the ones I've taken images from, in bold. Note that the only high-resolution footage I could find were the advertisements on youtube.
  1. "Gold Rush: The Mother Lode" (4.55) cnbc link
  2. "The King of Refineries" (4.28) cnbc link
  3. "Gold Mine" (1.56) cnbc link
  4. "Gold Rush: The Bullion Trail" (4.11) cnbc link
  5. High Definition Advert for the Series # 1 (0.31) youtube link
  6. High Definition Advert for the Series # 2 (0.38) youtube link

Clip #1 is the one referenced by the 2011 Zero Hedge article - the ZJ6752 bar is obviously the one most clearly shown since Bob flips it in front of the camera. The only commentator I know of looking for additional serial numbers was 'Conrad Murray' in the Zero Hedge comments section (link):

Conrad Murray says: "Frame at 1:31, 12 bars showing numbers.
2 are on the most recent bar list (didn't check all others)
I want to put right a misconception - those numbers shown are most likely NOT serial numbers (95% sure). Typically, the serial number markings are found on the top of the bar - which these days, is preferred for London Good Delivery bars (ref: Page 9 "Gold bars must be marked on the larger surface (the cast surface at the top of the mould) of the two main surfaces of the bar."). The bars in the shot above look very much like the US Mint 'bricks' with the five-digit number on the end, so we highly doubt those are Metalor bars.

This image is from a documentary on the Federal reserve. The shape and make
of the bars are a good match, but unfortunately the number appears to
indicate the 'batch' or something else, but it's not the serial number.
The US Mint bars are actually quite interesting - we catch sight of a few of those brick shapes in the Bank of England vault (professor and queen) tours. But that is another study for another time ... back to Bob Pisani ... unfortunately, the camera moves around too fast to spot the serial numbers on the other bars in the rest of the clip. Conrad speaks of another clue: "At 3:13 bar is weighed, 394.024."

The gross weight (oz) for one of the bars they are weighing (in video clip #1 @ 3:13).
That's a good start - in terms of investigation strategy. Barely satisfied however, I dug deeper. The weighing exercise is being referenced against a list seen at 3:22, where a guy is checking them off.

Video clip # 1 @ 3.22 - bar list of JMBRAM bars (Johnson Matthey Limited, Brampton Ontario)
Because we have a database, finding these numbers is easy - but first let's take a look at what's going on there, in detail. The column in the middle is the fineness/assay, the column on the far right is the fine weight (oz). If you look closely, you can see (2nd column) that the refiner brand is 'JBRAM' which also matches the logo we see on the bars being weighed. The serial number is to the far left (but can't be made out except that it has 6 digits). The pen marks indicate the bars which have a slightly different gross weight ... i.e. the ones with the ticks (e.g. 394.000 highlighted) are probably the exact scale weight, the ones with additional numbers are off by 0.025 of an ounce, and the guy has presumably written the weight as it appeared on their scales (e.g. 391.150 according to the list, but is apparently 391.175).

from earlier graphic, using Photoshop to rotate the image.

This is a very small variance in the weight (to visualize, think of one of those tiny 1/10oz coins, and split that into a four pieces), it also gives us a data indicator to look for later when comparing serial numbers. i.e. our new 'exact match' between vaults should allow for plus or minus 0.025 oz. Again, I digress. We have a bunch of data points we can match ... brand, weight, gross oz, fine oz and we know that the serial numbers are 6 digits in length (note, for this exercise, we will use their 'new' weight marked in pen).

Limit of reading error is large and many lines are unreadable, however I've got photoshop handy which lets me zoom and resize (plus I've lots of experience with the Google Captcha images). For the numbers which are readable I get TEN EXACT MATCHES from the sheet of paper, for bars held in GLD. Most of these show up in the GLD list on 5th August 2011, so this (presumably) sets an approximate date for when the filming took place.

The spreadsheet below contains the data and references for these 10 bars, including relevant links to our GLD document archive, in case you want to do any matches of your own. The bars identified start with "18####", which you can also see from the piece of paper. It's worth noting that these bars didn't stay registered in GLD for long - NINE disappear from the GLD registry by 25th August 2011, the tenth one is still there - JOHNSON MATTHEY BRAMPTON '187256' weighing 400.300 gross oz, fineness 9957. Verification: GLD.20130111.144338.Barlist.pdf (Page 1832). This doesn't mean these bars are not still in the vault, it simply means we don't know where they are presently.


So when we see them weighing and stacking the bars, these are/were real bars in GLD too. The mathematicians out there may be pondering the probability of a false match (which is always possible inside such a large sample set) ... someone else can work it out ... the ten bars were matched (with no duplicates) on a six-digit weight, 4-digit fineness and barserial (length 6). How many 6-digit JOHNSON MATTHEY BRAMPTON bars are in existence? In total, we have records for over 45,000 in the database (which includes other funds).

Clip #1 @ 3:26 - stacking the JMBRAM bars which were being weighed.
Although the full serial number cannot be made out, the first two digits
appears to be "18####', which correlates nicely with our found data.

So, did the other video clips also have some good bar pictures? Yes they did - but the matching had mixed results ... The three bars in the foreground below are clearly marked "ZH9925", "ZH9924" and "ZH9923" (Rand Refinery bars) - all three which appear briefly in the Metal Securities Limited Gold (ETF Securities) data during March 2011 through to May 2011, but they seem to disappear from the records in June 2011 and they haven't shown up anywhere since. You'll note that the 'year produced' is 2009, so in theory they have been in a vault for a while. If we assume that the footage was taken at the time of the CNBC visit, then this suggests the bars were still in the HSBC vault even after they appeared briefly on the ETF Securities list (fancy that) ... this is what we call 'dark bullion' (because it appears to be there but is not showing on any lists).

Clip #5 (0.02) high res Rand Refinery Serial numbers.
Made an appearance briefly in ETF Securities (London).
VERIFICATION: ETF.20110512.000000.WebsiteBarList2011.05.12.xls
(Worksheet "MSL Gold", Rows 4931-4933)

From here however, the trail goes dark. The rand refinery clip (Clip #2) has lots of highly-visible serial numbers, but there are no matches. I'm still going to show the image however, because we do have 323 records in the 'XP' series - most of which appear in GLD, but also a few in GoldMoney Zurich.

NO MATCH for these. Pity.

^^^ from the Rand Refinery visit, these bars are on a trolley. (Clip 5 @ 0:23)
What a shame we found Zero matches for these serials.

Not wanting to restrict myself just to CNBC video clips, the following images of LBMA 400oz Gold Bars have been forwarded from Bron Suchecki and Nick Laird, with my added commentary. Please note that many images were looked at, but these were the only ones I found direct matches for (serial number, brand, weight, fineness). For a long time I've been wanting to get an inside look into the vaults, and here it all is, in graphic detail. Folks, these bars have a story.

^^^ From BullionVault's image library - a Johnson Matthey Bar: serial 21592, fineness 9976 - has been with BullionVault London at least since May 2011 (my records started, and was still there in August 2012). Having this photo demonstrates my interest in this exercise - from the photo we now know that the bar was produced in "2004" (which probably also applies to some of the bars from that 'batch' as well). VERIFICATION: BGL.20120809.122504.NeilJGeddes.pdf (Page 4, Line 4)

^^^ Rand Refinery ZA7798 and ZA7796 (red highlighted) are currently in GLD (and has been since the start of my records in June 2011), the vault is HSBC London. The fineness of both these bars in the database is 9989. Similar to the bar above, now we know that the 'ZA' series was produced in 2009, which is useful since we have records for 1,962 'ZA' bars. The other ZA bars in the picture are unaccounted for, but in theory they are in the same physical vault as the ones we know about. I have no idea where or when the picture was taken, sorry. VERIFICATION: GLD.20130111.144338.Barlist.pdf (Page 1547 - roughly the middle of the page for both)

^^^ This is a Russian Bar (OJSC Krastvetmet, Krasnoyarsk) whose current location is with ETF Securities in London. We know from the database that it was listed as early as mid-Dec-2011 as 'A016707' and then February 2012 they adjusted the serial number record to '2011A016707' to reflect the year, so that checks out. The bar is still there. We also have 137 bars on record which match the (7 digit series). The serial number adjustment is in line with the LBMA best-practice requirements and is something we will have to build into the database structure when we are analysing vault movements. The word in the circle is 'Russia' in Russian. picture source: internet (ack. David Levenson). VERIFICATION: ETF.20130112.015448.WebsiteBarList2013.01.11.xls (Worksheet: "MSL Gold", row 11315)

Finally, a HUGE disappointment was the lack of matches from what otherwise should have been a jackpot of serial bar correlations. I recently purchased a new LBMA publication titled 'The London Good Delivery List - Building a Global Brand 1750-2010' (link), which has the distinguishing feature of a gold-bar photo shoot. It's quite a magnificent shot but unfortunately as far as 'matches with ETF-held bars', there are no correlations. However this is not to say that we haven't gained any information from the picture - we now have some good references for bar markings and also year of production (batch runs). I am considering framing the cover (the contents of the book is also excellent and useful for my research into historically accredited refiners).

^^^ Image borrowed (from their website) for illustration. The double A4 cover shows
76 assembled bars and dedicates a page to explaining their origins - most come from
the Bank of England vault. Unfortunately none of these appear to be from an ETF.

Result? A tiny bit closer to the real world that sits behind the abstract lists. Tungsten? These photos we've seen just only represent a tiny proportion of the number of the bars out there, so it's still possible - but the main problem is that as more details emerge through study, the conspiracy theories can only be taken with increasing amounts of salt. For example, these folk (talking about ZJ6752) were deadly serious at the time they did the audio. Real question is; how many more similar stories can also be solved through the knowledge of just a few additional facts?

p.s. I have used my archived documents as links so that the article links stay relevant. The normal links are GLD and ETF Securities, if you want to verify against the originals too. The renaming process I use for my archive is described here (and I do not alter the original source).

p.p.s. No Chinese-refined London Good Delivery bars have yet been found in the extended data set


Anonymous said...

Amazing work Warren.

S Roche said...

Interesting and I'm sure owners of allocated gold bars will start searching for their numbers as the word spreads. The gold re-hypothecation scandal starts here.

Also, does 1% count as a GLD Puke?

Warren James said...

Thanks guys.

Yes, anyone who holds title to a specific bar who wants to check whether the bar is present (or has been present) in the ETF system, should email me - bullionbars (at) The search process is quick, free, convenient and verifiable - but despite the standing appeal, the number of inquiries I've had is still currently zero so this indicates either investor's level of interest in the matter OR the relative famosity of Screwtape Files (I'm understandably torn over the interpretation).

@S Roche, I am hoping my data will reveal more of the GLD Puke mechanics - at least, this is one of my next areas of investigation.

Gotta admit though: I am somewhat concerned that these investigations all point in the same direction, i.e. knowing the details starts to reveal ordinary stories ... I think I preferred believing in the fairy tales.

The size of the database now sits at 49 million rows of data in the gold warehouse and 144 million for silver. Most of that is repeated information since GLD and SLV put out a list every day, but it represents a great resource for data mining.

Bron Suchecki said...

Good work. Finding photographed bars in the bars lists is great, sort of like seeing the bars in the vault for yourself, getting a sneak peak, and really starts to cut into the “the ETF don’t have any gold” meme.

The year pick ups are also good and will help with dark inventory estimates.

If people find PR images etc of bars on the net with legible bar numbers, suggest emailing them to Warren.

Re the reweights, note that LBMA weighing standards specify rounding of weights to the nearest 0.025 ounce. This means it is possible that the different weights are just different scales weighing the same bar but you only need a few fractions of ounces variance in accuracy one way or the other to get a different rounded weight.

The rounding rule seems a bit old school and out of date to me, I think based on the fact that most weighing was (and still is in a lot of cases) done on balance beams. You would think with modern technology they could get scale accuracy down to 0.001 of an ounce.

Did anyone try and contact Bob Pisani about the internet chatter that his visit was a fail for the ETFs? It would be interesting to get his first hand general impressions of how the metal in the vaults was stored and controlled.

costata said...

Great work Warren. Thank you for all of the effort you are exerting on this worthy project.

I'm wondering if this project could be commercialized. Could users anonymously insert their own allocated bar list numbers and receive an alert if any bars show up in your data when they shouldn't be there?

Even if the service has no commercial value (e.g. subscription) it might appeal as a gimmick to attract traffic to a site and be worth a fee from the site owner.

It might also be a way to finally monetize the silver and gold bug paranoia - Cash-4-Irrational-Fear.

Once you become famous you could then lead a campaign to crash a bullion bank. (JP Morgan is already taken of course but there are plenty of others to choose from.) Fame awaits you Sir.

Warren James said...

Thanks Bron & costata.

After watching all the videos carefully, I concluded that Pisani is actually a great presenter - and Bron, yes fully agree - I think the guy would make a great interview and I would love to hear his story and opinion on how his presentation was (ignorantly) misinterpreted by the 'bugs.

@costata, this does perhaps get to the heart of the matter ... no one wants to read a headline that says "Gold Documentary entirely validated - gold bars shown on camera match those found in publicly available vault records". Because my project is accidentally undoing the mystery, it will likely be selectively ignored by a large section of the very audience who drive most of the 'ogosphere. So sad!

re: website idea. It is an easy thing to build a website or side widgit which allows anyone to anonymously search for a bar number of their choice (a basic one could be set up within a few hours). The real limit regarding commercialisation is the level of interest, which is still a flat zero enquiries despite 'advertising' the option. I actually thought that investors would jump to the opportunity to prove or disprove a Level 3 argument (double counting bars) because they all bang on about it so much. Like I've said before, the absense of enquiries doesn't prove anything conclusive but I think it can be taken as partially indicative of how deep the problem is (i.e. not a problem).

As Bron suggests above, (open to anyone) please do send any picture you've encountered and I can analyse them (if the pics are good, and the bar story is interesting, I can even append them to this article - just be aware that I've already picked through a bunch of stock photography already, including google images, etc).

Bron Suchecki said...

I think the problem with a searchable list is that 1. it targets those with enough money to hold 400oz gold bars or 1000oz silver bars and 2. the subset of those who are untrusting enough who think BBs are doubling up would not trust a website to not store their enquiries/log cookies or somehow track them, assuming first you could convince them it wasn't a front for the US Govt wanting info on where all the gold is so they can confiscate it in the future etc etc.

Warren James said...

@Bron .. fully agree. I have toyed with the idea of publishing a static spreadsheet file containing every bar known and its current location, but there is no demand currently and no specific benefit to me (so even that won't happen in the near future). In the meantime I will continue to form interesting questions to ask the data :)

[p.s. Our bars document archive tallies the number of requests but the cloud service provider doesn't log details of requests, in case anyone is wondering.]

p.p.s. BullionVault is driving me nuts - the document url has changed again!

Anonymous said...

I could have helped last year as I had 10+ 1000oz Silver bars stored. I just got tired of all the "conspiracy" surrounding Silver and the joker Keiser. I have since sold them and gone "all in" on Gold (coins, allocated but no bars) after a laborious but fruitful study of FOFOA.

Keep up the good work.

Anonymous said...

There's a video of German gold available at the Bundesbank website. Perhaps it's of use for the project?

Warren James said...

@Börjesson, that's a very interesting (high resolution) video ... lots of 1968-produced Degussa bars, with some Rand Refineries bars on the bottom shelf in one shot. Thank you.

Interestingly, the database says there aren't any Degussa gold bars in the ETF's (there is lots of Degussa silver) ... which means we have another refinery whose output is not represented in the very large ETF repositories. This will be of interest to Costata who suspected that China's absence may not be unique. Of course I will double-check the data to back this statement ...

Looks like my next study will have to iterate the LBMA accredited gold refineries on the LGD list, which aren't present in the ETF bar storage.

Anonymous said...

In the same vein, there are three pictures of gold bars at the Riksbanken (Swedish Central Bank) website. They don't actually say that these bars are part of the Swedish reserve, but that seems a reasonable assumption - especially since they appear to be Swedish Boliden bars. (Note that "högupplöst" is Swedish for "high resolution".)

Warren James said...

@Börjesson, wow they are great pics, what a reference.

There are 1211 Boliden bars on record in the database, spread across these funds: "ETF Securities (x3), Goldmoney, SPDR, iShares Gold Trust". But, 99% of these are 5-digit serial numbers, ranging from 47066 to 54710. (The iShares Gold trust also records the prefix 'Nr.' as we see in most Boliden photos).

Boliden became LBMA accredited in 1984. I have a photograph record for a 1984-produced Boliden with serial "Nr37641", so (according to the range) the gold bars we see in the ETF's were produced primarily after 1984, there seems to be no ETF-occurence of the 4-digit serial range in the high-res. bank pictures from your link above (those bars are marked as being produced in 1955/1956).

Thanks for adding these to my library ;)

costata said...


I think the results of that data mining excercise you mentioned above in the comment about the Degussa bars would be very interesting.

As a cross check, perhaps Bron can mine the database of the Perth Mint. If, say, the LBMA has never supplied the PM with bars from the same suppliers then we may be detecting a selectively private market within the visible private market.

Anonymous said...


I'd like to try a hypothesis. Today, GLD's inventory declined from 1334.42 to 1332.61 tonnes. Looking at today's price chart, I am tempted to bet that this net loss of about 2 tonnes was in fact a gain of inventory and a loss of inventory of more than 8 tonnes, both during the same trading day.

After reading Bron's article, I am still a bit confused about whether the allocation would take place on the same day, but perhaps you fancy checking your data base for this signature today and over the next couple of days.


costata said...

Hi Warren,

Blogger appears to have taken a dislike to me and it has been putting my comments into the spam filter.

I recently made a comment about your exchange with Borjesson. I'm wondering if it is possible to identify any suppliers whose bars never appear in your data. The names may tell us something about the structure of this market.

Warren James said...

@costata - you are right (restored).. yes the blogger system seems to be over-zealous this week - possibly attributable to the rise of comments spam but it seems you have been unfairly targeted :p.

I agree - this kind of data will give us an overall picture - kind of like a fuzzy television reception. While some of the conclusions can be absolute, there are also elements we will never know and can't begin to speculate on because there is no data - for example the bar lists which will never be made public (like Fund of Canada) or confidential lists of privately held hoards.

But I will absolutely compile a list of expected vs. actual refinery representation - it's a start.

@victor, I will look into it - bear in mind that the serial number allocation appears to be completely arbitrary. The tonnage replacement probably won't reflect actual movements on a daily basis - i.e. we might only see 2 tonnes worth of bar serials change depending on their end-of-day routines. I do have some insights into their internal database structure but in a nutshell I don't think we can get calculations like that from the data. Aggregates over longer periods of time, we can (it's the subject of my next big article).

costata said...

Thanks Warren,

I think, with Bron's assistance, it will be possible to create a set of assumptions that can be tested using your data. This process of elimination will help with that.


Tony said...

I can only assume you've seen this. Nice clickable hi-res pic of some bars too:

costata said...


..for example the bar lists which will never be made public (like Fund of Canada)..

But we may still be able to gain insights that could lead to something clarification of the fuzzy image.

If, say, CBs and/or Treasuries in some jurisdictions have a mandate to exclusively source gold produced in their jurisdiction we could assume that all of the gold in their reserves is local.

Bron may be able to tell us, for example, if the Perth and/or the Australian Mints supplied all of Australia's reserves.

We know that US allies, such as Australia, sold down their gold reserves over a long period of time. We can accurately date the period when their reserves peaked and when they stopped selling down. If there are no bars of those "vintages" in any of the ETFs then it follows that those bars aren't circulating.

If any of these bars came into the market it could be instructive. The vintages of the bars in the ETFs could also be instructive. If there are no bars, or a statistically insignificant number of bars, beyond a certain vintage it implies that the stock is largely immobile as Another claimed.

I also think that this photo collection project could be expanded. I note that Ed Steer mentioned the bar list project in his latest newsletter. Perhaps some of these guys could put the word out that you are looking for pictures of bars with recognizable markings.

BTW are any of the ETF/ETP in your list based in Japan?

Warren James said...

@costata (just rescued some of your comments from the spam bucket - Google mustn't like your browser :)

You describe is a sensible framework for investigation. We do indeed intend to date the bars, using photos like these as clues.

Yes, Ed Steer's newsletter boosted the number of page views substantially (by nearly 700%, nice thanks Ed), but interestingly still no inquiries on 'can you look up this bar..' apart from @Börjesson's queries above. Increasingly I suspect that the folk who actually own 400oz babies don't waste their time with suspecting double-counting fraud, but that is still only speculation on my part.

There are quite a few funds that do not publish a public bar list. As such, I haven't been able to find any Japan-based ETF's, which does unfortunately mean that the Eastern hemisphere is grossly under-represented, with our closest data being Hong-Kong holdings.

For my part I am trying to build the bars index in such a way that it is transparent and reliable, which should form a solid base for any branch of research. I have certainly learnt a lot from the people I've talked to as part of this project, and likewise from the discussion in the comments. In such a manner I am more than compensated :)

My current area of investigation is to find an alternative explanation for the GLD Puke theory. At the minute I am learning the wise guidance from Kid Dynamite on the issue, but I absolutely would like to use the hard data to frame questions on Freegold theory as well.

Keep talking though, as I see some proofs emerging which could be applied against the data. Regards, Warren

Warren James said...

@Tony, thanks - that photo is a better quality one than the one I had. The numbers on the side of those 'bricks' (and the context of the photo) indicate they are US Mint bars - you can see the numbers are all the same and therefore they are all 'batched'. Interestingly, we've seen this stack a few times and it might be possible to cross-correlate the photos, but unless we get a close visual of the top face of the bar (with the actual serial number mark) then we won't be able to compare them against the ETF data (alas).

As part of our database structure we also hope to catalog the 'shape' of bars so that we can better narrow down a bar's origin, but we're not there quite yet ;0

Anonymous said...

Unfortunately I can't help with 400oz bar checks as we don't ship gold in from London - 300t+ a year from refining is enough for our needs.

Silver is different as our main silver supply is as a by-product of our gold refining so we do supplement by shipping silver in.

Anonymous said...

Not sure you have seen this video:!


Warren James said...

Hi VtC,

Thanks - a few others sent me that as well. I had a quick look at the first 15 minutes and got some good screen captures.

The bars shown are Johnson Matthey bars. This particular six-digit serial range is quite massive - in the ETF data we have on record about 55,000 gold bars of the possible 99,999.

With the bars in the clip, I can (unambiguously) make out serial numbers 163408,16314,163466 and 163497. These numbers don't appear in any of the Gold ETF data we have (about 2 years worth). What we do see is the sequence stop at 163394 then our records start again at 163500. It could be argued that the FED got the 1634XX batch.

Regards, Warren