Time to broach this subject again.
Just yesterday, the GLD saw a withdrawal of 8.88 metric tonnes. This followed a drawdown of nearly 23 tonnes on Wednesday. In fact, since the start of 2013, the GLD is now down 59.61 metric tonnes or 4.42% of "inventory".
Hmmm. Now where has that gold gone?
If the third bullet is true, then GLD drawdowns would be symptom of very strong, global physical demand.
So, for your consideration, let's revisit this issue. First, here's a reprint of the points that Andrew Maguire made initially. The full link can be found here: http://www.tfmetalsreport.com/blog/4327/guest-post-price-suppression-mecanics-gld-and-slv-andrew-maguire 
THE PRICE SUPPRESSION MECHANICS OF GLD & SLV
The bullion banks finance their ‘physical inventory’ by leasing it or selling it to GLD and SLV shareholders/investors, then the bullion banks in turn use these ETF’s inventories as a ‘flywheel’ to both manage and leverage their physical reserves. For this walk-through, I will use GLD as an example. (One can substitute SLV for all that is described below relating to GLD except the basket sizes are smaller, constituting 50,000 shares).
Baskets of GLD shares are bought and sold through a limited number of Authorised Participants. The authorised participants, (AP’s), are JPMorgan, Merrill Lynch, Morgan Stanley, Newedge (a joint venture between Société Générale and Credit Agricole CIB), RBC, Scotia Mocatta, UBS and Virtu Financial. This is how it is supposed to work. The size of each GLD basket comprises of 100,000 shares, each share representing just less than 1 troy oz. The AP’s, transfer ALLOCATED physical gold to the trustee who in turn creates the required number of new baskets of shares and then transfers these newly created shares back to the AP. To redeem the shares for physical gold or silver, the AP’s transfer any number of the baskets of 100,000 shares back to the trustee who then redeems these shares and transfers allocated gold back to the AP.
This is all well and good on the face of it, but there are a number of ways this ‘allocated’ gold backing the shares in the ETF can be diluted /hypothecated in order for the bullion banks to ‘manage’ their physical reserves.
If, as is often the case, there is insufficient allocated inventory available to the bullion bank at the current Comex driven & discounted spot fix price to create the necessary new GLD shares backed by allocated gold, then it is possible for a bullion bank to borrow short these GLD shares from the ETF instead of providing the required Allocated physical to the trustee to meet this obligation thereby ‘fly wheeling’ this physical demand in order to meet obligations elsewhere, likely at the day’s gold fix. This obviously has the effect of manipulating price lower vs. the true immediate supply demand fundamentals as no allocated physical metal has to be bought on the open market at that days fix to meet this new share demand as should be the case.
This is now the point where transparency evaporates. The AP claims to be Short GLD while concurrently claiming to be backing it with an equal size long ‘UNALLOCATED’ spot gold position. However, LBMA unallocated gold accounts are run upon a fractional reserve requirement and leveraged around 100/1 so there is very little need to back this transaction with any real physical at this point; this is left until later as explained below. To unwind this short GLD position, the bullion bank has to ALLOCATE the required amount of unallocated gold and then transfer this gold back to the trustee thereby receiving back the required # of shares in order to repay the original GLD shares sold short.
However, in conjunction with concurrent concentrated short futures positions, the sole object of this entire charade is to assist in depressing the price of gold at times of strong physical demand so that the futures price can be capped, usually at key inflection points where the price would break out and also swamp the very large concentrated Comex short positions. If this were not the case, the bullion bank would simply bid up that days fix price until it reflected that days true supply demand price levels for that fix and provide allocated gold to meet this real demand at that higher price.
The resulting distortion now created between the real and paper market price is exacerbated through the use of heavy position concentration and leverage in the futures and derivatives markets, where these very same bullion banks then seek to profitably repay the shorted GLD shares at a lower price at the point at or below where the lines cross profitably. This then puts these bullion banks in a position to finally spot index UNALLOCATED gold against this naked short position only then moving to buy the now discounted unallocated gold into the Comex contrived dips. These discounted unallocated long spot index positions are then ALLOCATED at the upcoming fix, enabling both the repayment of the GLD short position at a profit but most importantly controlling the rise in price against much larger derivative positions elsewhere.
Conversely, as evidenced by the steady 12-year stair step rise in prices easily observed in the daily and weekly charts, despite this many-year capping, we have also seen an ever larger and untenable LBMA unallocated short positions grow to what I now consider to be extreme danger levels. The reason is as follows: When the Bullion bank needs to make good on the unplanned/unanticipated CB and sovereign physical allocations at the fixes, they have regularly achieved this by going long GLD vs. short/selling UNALLOCATED gold. They then immediately turn around and transfer the required number of baskets of GLD shares to the trustee and receive ALLOCATED gold in return. Instead of settling/covering the short UNALLOCATED leg with this ALLOCATED gold, they are forced to satisfy these CB and Sovereign allocations by providing them this metal instead. The longer term price charts reveal this stair step higher, whereas we see no reduction, in fact from 2008 an increase, in the naked short Comex, (and unallocated OTC), bullion bank positions.
I hope this has been helpful in providing an insight into the internal dynamics of the ETFs and how the bullion banks continue to operate in the shadows.
Quite a few folks found this explanation a little too technical and slightly confusing. To help the cause, a few days later I took a stab at deciphering Andy's message:
Finally today, please allow me to take a stab at explaining in greater detail the "Guest Post" from Andrew Maguire. I posted it on Wednesday as we were leaving for Thanksgiving and I can see now where it caused some confusion. As you know, one of my favorite techniques for explanations is the chronological layout so let's give that a try. Additionally, I think I'm laying this out accurately. This is how I understand it. I'll check with Andy on Monday to ensure that this is at least close to being accurate. If it's not, I'll post some additional clarification then.
Anyway, I hope this helps explain the process. Again, the Bullion Bank that is also an AP of the GLD can "flywheel" metal into and out of the GLD and/or SLV anytime they need to in order to meet physical demand elsewhere. In the process, the BB/AP conveniently provides liquidity for GLD/SLV share demand, which negates additional GLD purchasing which would have otherwise been necessary. It's a true WIN-WIN-WIN for the BB/AP as they are able to cap and control price while appearing to have no problem meeting London demand and then they turn around and cover all the positions at a profit on the next bout of price weakness.
Again, THIS IS NOT SUPPOSED TO BE HOW IT WORKS. The banks are supposed to supply metal to both the GLD and the London buyers. There is not, however, sufficient supply to make this happen at the current price levels. So, instead of allowing price to rise to the natural equilibrium of buying and selling interest, the BB/AP uses the tricks outlined in Andy's guest post to manage and cap the situation. On the bright side, THIS CANNOT CONTINUE FOREVER and, WHEN it fails, the reset in price will be spectacular to behold.
By the time the next week rolled around, there was an active discussion on the internet regarding the accuracy of this analysis. (No doubt this post will reinitiate the "discussion" and bring out many of the same commentators.) Here's a link to the follow-up discussion, posted a few days later. Before you form an opinion on the matter, you'll definitely want to read both sides of the issue: http://www.tfmetalsreport.com/blog/4354/cage-match-bron-vs-denver-dave 
So there you have it. All of this should all make a very interesting reading assignment for you as we wait for today's GLD numbers. Could there be another huge drawdown? If so, what does it mean? Does it even matter? I look forward to reading your comments.
p.s. Andy just recorded this morning another interview with KWN. Be sure to check that site later today for the full interview.