the distillation process

[flyingdutchman]

Thanks Rad,
I really appreciative it.

[flyingdutchman]

Frost,
You guys are asking some great questions. Thats exactly what I hoped for when i started this. I wish i had all your answers but the truth is I dont.
1) Yes surface tension helps to prevent molecules from escaping which costs us energy (kw or propane). But does surface tension effect different size or type molecules differently I am not sure. If it does this could help our distillaion or hinder it depending on the molecule.
2) I strongly suspect surface tension helps in the formation of bubbles which on a plate is where I believe alot of interaction occurs between the liquid and the gas phase which is a good thing. maybe surface tension helps prevent weeping IDK (but perhaps appropriate vapor speed would be a better way to address that issue).
3) Maybe faoming in the downcomers could lead to flooding ( i think I read that somewhere) if so surface tension could aid in the foaming so thats bad.
4) Can you modulate surface tension, probably but it might require adding something to the wash (like surfacant) but I dont know if the end result would be good or bad.
5) IDK if plate shape will effect it but I could be way wrong on that.
Questions like yours are why we have the internet and exactly why people write scientific papers on research. Heres a trick with reading scientific papers. just read the introduction and the conclusions everything in the middle is often them saying look how i did it or proved it (LOL)

Here is a list of things that others say can go wrong with plated columns. I am not saying they are bad just that it is good to run them the right way to prevent these problems.

Vapour Flow Conditions
Adverse vapour flow conditions can cause
Foaming
Foaming refers to the expansion of liquid due to passage of vapour or gas. Although it provides high interfacial liquid-vapour contact, excessive foaming often leads to liquid buildup on trays. In some cases, foaming may be so bad that the foam mixes with liquid on the tray above. Whether foaming will occur depends primarily on physical properties of the liquid mixtures, but is sometimes due to tray designs and condition. Whatever the cause, separation efficiency is always reduced.
Entrainment
Entrainment refers to the liquid carried by vapour up to the tray above and is again caused by high vapour flow rates. It is detrimental because tray efficiency is reduced: lower volatile material is carried to a plate holding liquid of higher volatility. It could also contaminate high purity distillate. Excessive entrainment can lead to flooding.
Weeping/Dumping
This phenomenon is caused by low vapour flow. The pressure exerted by the vapour is insufficient to hold up the liquid on the tray. Therefore, liquid starts to leak through perforations. Excessive weeping will lead to dumping. That is the liquid on all trays will crash (dump) through to the base of the column (via a domino effect) and the column will have to be re-started. Weeping is indicated by a sharp pressure drop in the column and reduced separation efficiency.
Flooding
Flooding is brought about by excessive vapour flow, causing liquid to be entrained in the vapour up the column. The increased pressure from excessive vapour also backs up the liquid in the downcomer, causing an increase in liquid holdup on the plate above. Depending on the degree of flooding, the maximum capacity of the column may be severely reduced. Flooding is detected by sharp increases in column differential pressure and significant decrease in separation efficiency.
Column Diameter
Most of the above factors that affect column operation is due to vapour flow conditions: either excessive or too low. Vapour flow velocity is dependent on column diameter. Weeping determines the minimum vapour flow required while flooding determines the maximum vapour flow allowed, hence column capacity. Thus, if the column diameter is not sized properly, the column will not perform well. Not only will operational problems occur, the desired separation duties may not be achieved.

[flyingdutchman]

I wasnt going to get into this but others have asked about intermolecular forces. I think there are generally 2 kinds that we might call van der waals forces (weaker ones) and hydrogen bonding (stronger).
here is a link that explains them http://www.chemguide.co.uk/atoms/bonding/hbond.html#top go to the second page for hydrogen bonding or herehttp://www.chemguide.co.uk/atoms/bonding/hbond.html#top

Note that alot of our things on the list of subatnces in the wash can and do easily hydrogen bond to each other (esp with water) and that it is a strong interaction.

A hydrogen bond is the attractive interaction of a hydrogen atom with an electronegative atom, such as nitrogen, oxygen or fluorine, that comes from another molecule or chemical group. The hydrogen has a polar bonding to another electronegative atom to create the bond. These bonds can occur between molecules (intermolecularly), or within different parts of a single molecule (intramolecularly).[2] The hydrogen bond (5 to 30 kJ/mole) is stronger than a van der Waals interaction, but weaker than covalent or ionic bonds. This type of bond occurs in both inorganic molecules such as water and organic molecules like DNA.
Intermolecular hydrogen bonding is responsible for the high boiling point of water (100 °C) compared to the other group 16 hydrides that have no hydrogen bonds. Intramolecular hydrogen bonding is partly responsible for the secondary, tertiary, and quaternary structures of proteins and nucleic acids.

[flyingdutchman]

So if you understand that then me show you this...

This is why is am very hestitant go along with the idea that water and ethanol separate on a plate at elevated temperatures i am not saying its impossible I just am extremely doubtful. I encourage anyone who does think so to find any scientific literature that describes this. It would go a long way towards making a believer out of me.

Solute-Solvent Interactions
1) Miscible liquids: mix in any proportions.
2) Immiscible liquids: do not mix.
3) Soluble liquids: mix in certain proportions. (in general we heat the mixtures in order to increase solubility and if anything shake sonicate and vortex)
4) Intermolecular forces are important: water and ethanol are miscible because the broken hydrogen bonds in both pure liquids are re-established in the mixture (infact they are often described as infinitely miscible).
5) The number of carbon atoms in a chain affect solubility: the more C atoms the less soluble in water.
Here is an image that shows hydrogen bonding (images say so much more than words)

So if we add heat and break the H bonds in the miscible mixture they will just reform. If we continue adding so much heat that we break all the H bonds and keep them broken then I suspect water is not H bonded to water and ethanol is not H bonded to ethanol so nothing holds either pure liquid/substance together. Anything that doesnt become a gas I suspect is randomly bouncing around which is a big difference from two ordered pure liquids (once described as completely miscible) separating from each other so that the denser one sinks to the bottom. Again I am not saying its impossible but after ALOT of searching i find no literature that suggests this occurs (lets not even get into mixing and entropy).
Infact in nthe case of phenol (another alcohol but greasier or more carbons in a ring) and water I believe they report increased temperatures increases solubility and not the reverse.

see above 70 it become a homogeneous solution sort of the exact opposite of it separating out.
Please remeber my point here isnt to say anyone is wrong or right its just to illustrate why I think this way. I would never say anyone is dead wrong because alot of people have made that exact statement only to have to eat those words later.

[Frosteecat]

So we should just distill through agitation and freezing and stop all this silly packing and heating talk!!

[LWTCS]

Good information indeed. But within the context of what constitutes a good drop,,,,I'm not so sure that we need to worry that solubility for minute constituents is increased? Unless we are after perfectly pure?
Please continue....This info is very good.

The example I was trying to make with my stratification comment was merely that the most volatile constituents must occupy the highest space when an optimal amount of heat is applied...
Mixing in my particular case gets largely mitigated as I can completely isolate several bodys of liquid and yet recycle till the lions share of the most voilitile constituents occupy the highest available space...
Entrainment may be an issue for some with this approach? I rather enjoy my rum the way I smear it (contrary to some earlier comments).

[flyingdutchman]

Its a great idea dont get me wrong. If we could get the most volatile liquids to rise to the top it would be perfect. its just like you guys said the most volitile component would be right at the interface between gas and liquid. The perfact place to interact with the vapor and an increased local concentration effect. I personally think it would be a huge boon. But to do the you have to break all those hydrogen bonds that we just spoke about.
1) Hydrogen bonds do things like hold your DNA together they are really strong in the world of intermolecular forces.
2) think of hydrogen bonds more like magnets not sticks. When you break them they just reform right away and there are lots of them like velcro the sum of the individuals add up
3) Plus they arent just soluble liquids they are miscible which is a whole step above soluble. Plus they arent just misicble liquids they are infinitly missible liquids. when people talk about miscible water and ethanol often come up just because they are so miscible.
See infintely miscible and stratified in the same solution are like a contradiction of terms. Sure if we apply and keep applying enough energy we break H bonds and van der waals interactions, but I dont think you do this selectively. So now if all the bonds are broken what force is there that holds water to water or ethanol to ethanol. Now the term entropy dominates (random disorder, the world wants to fall apart, like diffusion) and what I see in my head is like a bunch of pool balls during the break. Its not solids at one end and stripes at the other or a stratification of stripes and solids from one end to the other, they are just bouncing all over the place. Add to this that a bubble or sieve plate is mixing constantly just by the nature of how they operate (remeber we stir shake and vortex to cause mixing) and it increases mmy skepticism. If we could stratify water and ethanol I honestly think you would see it reported with quantified measuarble values in like the CRC handbook of chemistry and physics or in some ACS paper describing the variable factors in distillation. look at the reported example of phenol and its just partially soluble. The graph shows the more they heat it the more soluble it becomes.
Sorry LWTC I am not neccessarily right and its is entirely possible that its exactly as you describe it, thats just how I see it.
See working in the sciences i am natural born cynic any idea is a good one but until some shows me solid, quantifiable, repeatable evidence that is beyond the margin of error I leave it as a cool idea but not proven.
quod erat demonstrandum

[LWTCS]

Sorry LWTC I am not neccessarily right and its is entirely possible that its exactly as you describe it, thats just how I see it.

I think your dead on point....
I am merely trying to create a way that allows me to understand and illustrate in my minds eye,,a few things that I have visually observed and also help me understand why something is happening with these deep liquid beds that some folks said could not ... if that makes any sense?

Again, it seem clear to me based on your lesson that "stratification" is the wrong descriptor.

[LWTCS]

Here is another illustration that I often think about with regard to how this system works.....Hope it doesn't seem to odd.....
I'll start by saying that I'm not so sure that breaking hydrogen bonds to the enth degree is needed for our purposes.

Think how a painter might clean his brush..
The brush gets dipped and agitated into a solvent that gets dirty very quickly. The solvent gets dirty and the brush is not entirely clean. Perhaps some of the solids may have been reintroduced back to the brush during agitation?
So then the brush gets dipped and agitated into a new bucket of clean solvent..More solids drop out of the brush.The bucket become less soiled than the previous bucket and the bush becomes that much more clean. And perhaps some solids get reintroduced during agitation?
Then the brush gets dipped into a third bucket of solvent. The third bucket acquires a slight cloudy appearance and the brush appears to be generally clean. Perhaps a small amount of solids got reintroduced during agitation?
So in order to make sure our brush is good and clean we dip and agitate in one final bucket of clean solvent.
As a painter knows, often times the best solvent to use for cleaning the solids out of one's brush,,,, is the very solvent that is in one's paint.....Ethanol is a solvent BTW.

Seems a bit simplistic I know?...I can't do the math,,,but I know how to clean a brush with solvent.

[Frosteecat]

That is a great analogy and certainly held up by the action of packing, plates and thumpers. If a high concentration of "solvent" is effective in screening subsequent volatile mixes, I would assume that the goals we are striving for would make the use of staggered high ABV filter systems---whichever is most effective or a series of mixed methods that are synergistic---the best solution. Being able to control them interdependently allows us to influence ABV, flavor, etc. But the "wildcards" are thermodynamics, pressure, vapor flow and so forth, right? Much like multi-column distillation practices, we need to figure out the best hardware to fuel to mix methods and play them like virtuosos during our runs. Segmentations of some sort, and using different components that aid each others' efforts in the right sequence seems paramount to solving some of these small scale attempts. I see packed thumper and inline condensers, reboilers & packing/plate combinations being the solution...but in what positions/sizes and at what temps/reflux rate the mystery.

I'm a bit unclear as to what bubbling is doing to aid the process...does it aid in separation due to the added volatility of the wash? If so, wouldn't we want a series of very volatile bubbling segments (bubble plates or "chambers"?) in series with thumpers?

[Frosteecat]

So that's why many plates are preferred...but we are showing that packing is also quite effective...why don't we have an incredibly volatile chamber or surface apparatus, into packing to send the lesser fluid back through the bubbling area, and then perhaps into another similar bubbler/packing segment? Why are we limited to a single 2-4" plate with 3 or 4 bubblers? Can't we have a large bubble chamber with some significant head space into a packing section?

[Prairiepiss]

Frosty this mite help you a bit?

And I home you don't mind I copied this from the seperation or phase change or mass transfe thread.
http://homedistiller.org/forum/viewtopi ... e#p6992377

My understanding of it is.

Mass transfer. The movement of the different compounds within the column. Whether they are being carried up the column within the vapor movement. Or being carried down within the liquid draining down the column.

Phase change. Would be the changing from one phase (vapor) to another phase (liquid). Or visey virsey. From liquid to vapor.

Separation. This would be the actual separation of the different compounds. And is carried out during the phase change and mass transfer operations. It takes both mass transfer and phase change to get separation.

Basically you have a vapor rising carrying alcohol and water with it.(mass transfer) It goes through a phase change from vapor to liquid. Then the liquid drains back down the column. (mass transfer) And again as the vapors rise through the liquid the water in the vapor does a phase change to liquid and drains via mass transfer of liquid. The alcohol in the falling liquid phase changes to vapor again and rises VIA mass transfer of vapors. So you get separation alcohol traveling up VIA vapor mass transfer and water travels down VIA liquid mass transfer.

So phase change within mass transfer = separation.

[Frosteecat]

Yes that helps but I guess I'm wondering why I'm not seeing a larger and more volatile and longer device/chamber that aids the process? Are we limited by the bubbling and change occurring due to vapor pressure (I.e. through an upcomer or perf ONLY forced by temp and pressure) or can we help make it more volatile mechanically (an "agitator" built into the system somewhere) or even by restricting vapor flow strategically as it enters an agitation segment/phase? I know it's only a piece of the puzzle but seems like it might multiply separation effect in a smaller/shorter/easier fashion...

[LWTCS]

Yes that helps but I guess I'm wondering why I'm not seeing a larger and more volatile and longer device/chamber that aids the process? Are we limited by the bubbling and change occurring due to vapor pressure (I.e. through an upcomer or perf ONLY forced by temp and pressure) or can we help make it more volatile mechanically (an "agitator" built into the system somewhere) or even by restricting vapor flow strategically as it enters an agitation segment/phase? I know it's only a piece of the puzzle but seems like it might multiply separation effect in a smaller/shorter/easier fashion...

Your questions are so good frosty...
Though this is counter to my poorly chosen "Strato" example...Good vapor speed can assist in creating better mixing...though not so much that it would create blow by as demonstrated in Harry"s column vid.....
We feel that the smaller bath volumes are limited as that plate represents only a single distillation compared to many distillations within a comparable space of structured packing...That's why the texts say you need twelve (or so ) plates to get azeotrope.
But if you recycle (via good drain circuitry), draw no product and facilitate a larger amount of alcohol in the apparatus then in effect the apparatus gets more efficient as the equalization period progresses.

[Frosteecat]

Ok I get that...just want to optimize it!! Can't wait to get home and have the house to myself for a change tonight, pour a couple tall ones and luxuriously read through several of the theory and design threads that are active right now! I have discovered that my brain seems to require ALL information about a subject before the concept locks into place...it can be maddening for myself and those around me, but it is why I am leapfrogging one step at a time through all this...
This is the perfect environment for me and I am SO thankful I have access to the practical and professional knowledge base you guys can tap into for me... As long as you're willing to answer, I'll keep asking!!

[flyingdutchman]

Ok,
Sorry for not getting back sooner. Went out bass fishing yesterday (best 2-6 largemouth, 23 for boat total) then came home and over achieved with my own product so running a little slow today. you guys are miles ahead of this but I figured I would just run thorugh a brief explanation of pg 2 for anyone who reads this in the future.
This Diagram is a visual explanation of what happens in the boiling condensation cycles of a IDEAL mixture of 2 pure substances (A & B) and is read from right to left. Lets sat A is water with is higher boiling point and B is ethanol. Some important points are...

BP mole fractions.gif (5.36 KiB) Viewed 2004 times

1) The y axis is B.P.
2) the X axis is mole fraction but just think of it as what percentage/ratio we have of A vs B. as we move right to left we go from pure A to a mixture of more and more B in A until we get to pure B
3) We start with a mixture of with a small amount of Ethanol in Water at point C1. How much is shown on the X axis.
4) when we boil it we follow the line up fron C1 to the blue liquid composition line then across to the pink vapor composition line.
%) Because like we said everything evaporates BUT not in the same ratio, the new vapor has a greater ratio of the more volatile component EtOH (ethanol) and less of the less volatile component H2O.
5) How much more ethanol is shown when we follow the light blue line downwards to point C2 on the X axis. See it will contain a grteater percentage of EtOH. Every time we repeat this stepwise the percentage of EtOH increases. Just follow from C1 to C2 to C3.
6) Equally important is what happens to the boiling point from C1->C3. The BP of C1 is greater than>C2 and the BP of C2> C3. So as we get closer to pure A the BP of the different mixtures get closer and closer to pure A BP.
Now lets look at this from the inside of a column. Lets say the boiler (not shown) is P1 then the lowest plate is P2 and the upper plate is P3.

bubblecaps.gif (4.5 KiB) Viewed 2004 times

1) So the compostion of the mixture in the boiler P1 is C1. When we boil it we get a new vapor who we know has a compostion of more EtOH.
2) How much energy it took to boil it is determined by the Latent Heat of Vaporization. On a gram per gram basis this is different for different substance but on mole per mole basis the energy is close enough to the same that we dont need to worry ourselvers about the difference for what we work with.
3) Note that the LHV is alot of energy and for water i thinks its 5xs more energy to go from liquid to steam than it takes to it 1 degree celcius.
3) the vapor rises upwards it velocity is determined by how much vapor we are creating 9heat input) vs the column diameter (because gas expands to a greter volume than the liquid.
3) As the vapor is created at P1 it actually has composition C2.
3) When this vapor rises upwards it meets the liquid layer on P2 also with the compositon C2 and condenses giving up its energy and in the process boiling the liquid on P2.
4) I think the signficance of the bubble beyond the fact that this is where the liquids and vapor equilbrate is that bubbles have more surface area so more area for liquid and vapor to interact (thanks to surface tension).
5) the new vapor created at P2 has compostion C3, it rises up to P3, where oit meets the liquid on P3 of composition C3 causing it to boils and creating yet a new vapor I will call C4 and represented by the green line on the top diagram.
6) should any liquid leak from plate P3 down to P2 I believe it would enrich the composition of P2 beyond C2 increasing its percent ethanol
7) If we continue to condense a large portion of the distillate at the top of the column and let it return to P3 which overflows down to P2 then I suspect we will continue to enrich compositons C2 and C3 (with greater and greater percentages of ethanol) as if we were actually adding plates (even though physically we are not)
8- Ofcourse at some point we actually want to draw off the distillate so this is a kind of balancing act.

Others who understand refluxing stills better than me can answer this part better than me or even correct me here but I think this is the part that would give us heads compression. Mostly because by using a high reflux ratio I believe we can enrich the vapor with more and more of the cogeneors that are more volatile than EtOH in essnce compressing the volume of contaminated EtOH.

[LWTCS]

That sounds spot on to me FDM...

[flyingdutchman]

Cool thanks.
PS
LWTC if you say your getting better separations than you should be your word is solid with me, Im a believer. There absolutely has to be a reason. Maybe first thing is to look at your design again and figure exatly how many plates that should add up to. I am guessing that is already done but do you have a link or better yet a drawing so everyone can see it at once. Also I forget (CRS) i dont think there was a dedicated refluxing coil at the top but am i wrong?

[LWTCS]

Maybe first thing is to look at your design again and figure exatly how many plates that should add up to.

4 plates not including the boiler or dephlegmater.

the liquid volume in the largest thumper is actually 7 liters not 2.5

[Frosteecat]

Ok...now THAT looks like what my gut is telling me! I AM NOT AN IDIOT! YAAAAY!
Ahem...sorry.

[LWTCS]

Ok...now THAT looks like what my gut is telling me! I AM NOT AN IDIOT! YAAAAY!
Ahem...sorry.

So much of what you have been thinking is similarly contained within this concept. I didn't realize you had not seen what you have been thinking.

[Frosteecat]

No I had not...this was just what my intuition was telling me. Of course I'd want to add mechanical bubblers, pinwheels and centirfuges too... ... but the "straight shot, simple plate or packing" clan was pissing me off cause my mind kept saying "no no no you need combos and rest stops and multiple effect areas with traps" etc.
It's a nice validation but I think you can do more and better.

[LWTCS]

The dynamic boost happens when the top plate starts drain feeding the middle plate,,, as the middle plate continues to feed vapor into the top plate....

Once all three chambers are cycling and no product is collected,,,the most volitile material must occupy the highest space. When that space is exhausted, the highest voilitiles have no choice but to escape as liquid and drain down to enrich the next lower plate level. This continues until the entire tower section is stacked with very proofy liquid....

This deep body of very proofy liquid allows for a very fast collection speed. Not at all like maintaining yer collection cup in a Bok. At this point not much heat input is needed at all to drive the system as the boiler temps naturally rise.

Never really felt smart enough to try and mitigate any entrainment issues......This is after all my rum maker and I like the way I been smearing it up
I pull heads old school before I put her into the big reflux cycle.

[Frosteecat]

Well, I'm a big believer in turning liabilities into assets so I simply seek to find out how things want to behave and then give them "choices" or sometimes impose a choice that they don't know they're making with my influence.
Works great with kids too...

[flyingdutchman]

Hi LWTC,
Hands down no questions asked, I have to say that is a very original design. I looks like you have a huge area for vapor to interact with liquid and I bet it can piss like a race horse when you out the spurs to it. I suspect trying to calculate the number of theoretical plates you are getting would be a whole lot easier (and more accurate) with the dephleg completely bypassed (ie no refluxing). It looks like 5 condenstaion cycles (including the primary boiler) if the dephleg is bypassed. Have you ever run it with the dephleg bypassed. if you did how many theoretical plates would account for the purity you are getting in the parrot vs starting wash % alcohol.

[LWTCS]

No Dephleg is 87% and that would be with a water precharge at each plate to insure that phase change is initiated at every plate level.

[Frosteecat]

Maybe a dumb question but why wouldn't you have the option built in to return all of your distillate at various stages to all of the lower chambers in any sequence or at any time (accounting for flooding,etc of course)? Wouldn't you boost ABV by returning "tops" to the bottom or "upper bottom" as desired?

[flyingdutchman]

Frost,
Good idea, and it kind of goes along the lines of Myles current thread on reflux splitting.
LWTC,
Mike Nixon wrote a paper where he talks about a Vapor Reflux System. In it he advocates returning the reflux to essentially what is your parasitic/secondary boiler. Probably you guys already know all about this and maybe its already been talked about in Myles thread i really need to check that one out.
In a way judging from your drawing i wonder if you are sort of indirectly doing this already. I mean you havent directly routed your reflux to the seconday boiler BUT by the nature of what to me looks like an overflow pipng system (to prevent flooding?) I think you might be doing it indirectly.
PS
Under reflux I would have a hard time saying exactly how many plates you should be getting. Maybe Rad or PP has a formula to calculate this. However I believe the most accurate way of checking how many theoretical plates your are achieving strictly from your inline deep bed liquid plates is to run it with no reflux. see what ABV it comes off as, compare that to the ABV of the wash and use the interactive calculator on the parent site to tell you exactly how many "plates" you have achieved.

[flyingdutchman]

You guys are probably gonna laugh your sox off but...
After all this i have to admit i am a pot stiller LOL

PS
LWTC I am not so sure you have entrainment issues. I mean, I think entrainment to strongly related to vapor velocity. Velocity is related to column diameter but your case is kinda atypical. The vapor velocity at your "plates" (which I think is the important part) is related to Thumper diameter and not the diameter of the column connecting them. What is the diameter of your thumpers because if I remember right its like 4 inches minimum which means it has four times the volume of a 2 inch column and that would be a whole lot less vapor velocity than a 2 inch column. Atleast thats what I think???

[LWTCS]

I started running this rig before Mr. Nixon ,,(or Harry rather) posted that paper....I was observing things that seemed to be quite contradictory to what I was hearing from the very smart fellows...so I was very confused..How could all those very smart fellows be so off point in key areas?

When I read Mr. Nixon's paper I was inspired and at the same time things became much more clear to me as I had made observations that Mr. Nixon referenced.....Mr. Nixon does an excellent job with his narrative to keep a journeyman like me up to speed,,,I felt any way?
My input on the thread was dismissed and was told that my rig had nothing to do with
Mr. Nixon's "vapor reflux" concept.
But to me the concept is best illustrated on Mr. Nixon's paper when he discusses "Positive Feedback".
The notion of positive feed back is the hallmark of what I am doing.

Seems folk could not make the connection for a couple different reasons?...My inability to effectively explain the similarity of the concept to what I was doing ,,,,my equipment was clearly a different design and also when I use the word "thumper" it threw the folks off target,,,as every body knows that a "thumper" is pretty stone age technology....

You should read the Paper (and the thread). And think more specifically about "positive Feedback" after you read his paper...