Date   

Re: The G-N Distribution Model

Aley, Jeff A
 

George,

 

              I think your message achieved its goal – it was entertaining.  I see a couple of different philosophies.  In the one case, a person could model the railroad exactly as it was and exactly as it operated in a given place and time.  I think very few people (none?) actually do this, as the operations would be totally prescribed and therefore very boring.

              A different philosophy is to “kinda” model a specific time and place, but leave some leeway for us to play with our trains.  The nature of most op sessions means that we are creating our own approximation of what “could have happened” and not “what actually did happen”.  And it is within that approximation that guidelines like the G-N theory become helpful to people like me.

 

Regards,

 

-Jeff

 

 

From: STMFC@... [mailto:STMFC@...]
Sent: Tuesday, September 05, 2017 5:47 PM
To: STMFC@...
Subject: [STMFC] Re: The G-N Distribution Model

 

 

I post to change no one's opinon, merely to hopefully entertain.    I model s.w. Virgina.  Ton of Southern and Interstate hoppers.  Now to model a single day in reality and follow photos, regardless of the G-N distribution model, there is one day when I have to model an empty Lehigh Valley hopper at a s.w. Virginia mine.  Thus there is an inherent philosophical conflict between modeling reality on one day and distribution models.

  Should one only model undamaged cars because they represent the highest range of freight cars on any given day?  Or model a damaged car you see on a given day?  I watched too much football this weekend and am trying to distract myself.  Still if you follow a distribution model does that philosophically mean you must model an extended time period rather than a single or small time period?  Or if you model a specific time or a single day, then any average or norm must be out the window?

 

George Courtney

 


Re: Digest Number 11124

Brian Termunde
 

Blaine,
First I think it's great that there will a second RPM out here in the Intermountain West. Second, Count me OUT!

Why? Well mostly since I work Saturdays, and with the less then 80 hour notice, it makes it nearly impossible to arrange a swap or anything! As a constructive criticism, wasn't possible to announce this event further out?

It sounds like a great event, and as I did the first meet, I would love to attend and support this one, but I am doubtful that I can rearrange my, and someone elses schedule on such SHORT notice.

Take Care,
 
Brian R. Termunde
Midvale, Utah


-----Original Message-----
From: STMFC
To: STMFC
Sent: Tue, Sep 5, 2017 4:57 pm
Subject: [STMFC] Digest Number 11124

There are 15 messages in this issue.

Topics in this digest:



5. Mountain States RPM
From: blainehadfield


Messages
________________________________________________________________________

5. Mountain States RPM
Posted by: blainehadfield@... blainehadfield
Date: Tue Sep 5, 2017 1:50 pm ((PDT))

Railroad prototype modeler meets have finally come to the Intermountain Area! This Saturday! 

Connect with friends and fellow prototype modelers at the only prototype modeling meet in the Intermountain West. Enrich your skills by interacting over detailed models and content-driven presentations. Nationally recognized historians and modelers. Clinicians from Washington, California and Utah: Rick Selby will present on Per Diem boxcars; Brian Rutherford will present 'Modeling from Prototype Inspiration: The WP and the Inside Gateway.' Kevin Packard will share tips and tricks for photorealistic weathering with a live presentation. Jere Ingram will present an introduction to modular signal systems. Blaine Hadfield will give an executive overview of ExactRail from its origins to the present. 

Railfan the historic trans-con from the machine shop and roundhouse. Shop at Spring Creek Trains. Meet with Canon & Company's David Hussey and see his amazing line of parts and boxcar kits. The Union Pacific Historical Society will be in attendance, and Scale Trains will have a display table of future releases. Enjoy our barbecue on the patio. Raffle Prizes from Intermountain, Kadee, ScaleTrains, ExactRail, and Tangent. Fellowship. 

Bring your models! There is no cost to display. There are no contests. 

Mountain States RPM is at the historic Evanston Roundhouse & Machine Shop (1500 Main st.) in Evanston Wyoming. Doors open at 9:00 a.m. on Saturday, September 9th. The raffle begins at 5:15 p.m. See our Facebook page Mountain States RPM for more information. 

Admission is $20.00. Kids under 13 are free with paid adult admission. 

We can't wait to see you there! 

Blaine Hadfield, 
Mountain States RPM 


Re: The G-N Distribution Model

Tim O'Connor
 


Rich, exactly! In the software I wrote many years ago, I had a "dwell time"
value for cars that left the layout via staging. This allowed you to VIRTUALIZE
your fleet so that one PRR car for example could represent 100 cars of a fleet
compared to one Tennessee Central box car. In this scheme, the random arrivals
of cars can produce an AVERAGE of PRR cars being 100 times more likely than the
TC car. And this virtualization of the fleet can be accomplished with 1, 2 or
100 PRR cars - depending on your modeling tastes. The number of PHYSICAL cars is
simply the upper limit for the random arrivals of that fleet. And each physical
cars gets it own individual dwell time value, so you can virtualize the types
of cars for a large railroad like the PRR, so an X54 is much less likely than
an X29 for example.

What I had in mind was cars like depressed center flat cars, or whatever, that
are uncommon but are certainly seen now and then. But it worked just as well for
fleet virtualization.

Tim O'Connor



And you can carry the reality even further by having off staging storage for those cars that might only show up once a year.

For example, the Elkland Leather tank car carrying spent tanic acid may only move to a processing facility once every 3 or 4 months in reality.  So the car when returned empty is taken off the layout and stored for the next Y number of operating sessions.  You determine the Y.  In the case of this car and operating twice a month it would sit out for 6 - 8 sessions.  All it requires is another car card that indicates when it is to be cycled onto the layout.

I use Elkland leather as my example because I am currently seeking information on these cars to build one of their tank cars which will appear perhaps only once every 6 months.  The planned fleet for the Western Pennsylvania Model Railroad Museum has a number of these cars which all are cycled following the same method.

Rich Orr


Re: The G-N Distribution Model

Tim O'Connor
 


 Dave Evans wrote
 (For your coin toss analogy, the odds of tossing 50 heads in a row is one probability.
 The odds of throwing 50 heads in a row out of 100 coin tosses, that totaled 50 heads and
 50 tails, is much less likely - i.e. tossing 50 heads in a row, along with two streaks of
 tails in a row that totaled 50.)


Sigh. Another one who has no understanding of probability and statistics.

100 coin tosses yields 2**100 (2 multiplied by itself 100 times) or

  12,676,506,002,282,294,401,496,703,205,376

possible order-dependent outcomes. EACH of those outcomes is equally likely.

The AVERAGE of all possible outcomes is ... 50 heads, and 50 tails.

However, we don't care about order. We only care about the number of heads and
tails. That turns out to be obvious - there are only 101 possibilities of heads
and tails in 100 tosses - from 0 to 100 heads.

But now ask, what is the PROBABILITY that there are 12 heads? That turns out to
be vastly more complicated, because that question is really asking about how many
of the 32 digit number of outcomes above, have exactly 12 heads?

Using the formula for a binomial coefficient, written as (N over K)

     N!
 -----------
 K! * (N-K!)

Multiply this coefficient by the binomial probability, which is:

 [(p^k)]*[q^(n-k)], where p = probability of success,
                         
q = 1 - p,
                         
n = number of trials,
                         
k = number of successes

The general formula for binomial probability:

            
( n )   
  B(n,k) =   (   ) * p^k * q^(n-k)
            
( k )  

Instead of doing all the math yourself, I recommended this handy calculator http://vassarstats.net/textbook/ch5apx.html

You may be interested to know that the probability of exactly 50 heads and 50 tails
is not 50 percent - it is in fact, just under 8 percent. I leave this as an exercise
to the reader (using the calculator).

Tim O'



Re: The G-N Distribution Model

devansprr
 

George,

If you are only modeling a small quantity of cars, the probability that the car you modeled (assuming it is a prototype) would appear at that location, on that day, is about the same for any car, if you knew that at some point an LV car showed up MT at a s.w. Virginia coal mine. At that instant, the probabilities for any hopper that ever was loaded at that mine, appearing at that instant, are the same. But if no one has ever seen an ATSF hopper at the mine, then using that car is a problem.

As Dennis noted, if you are only going to model one day, then you have lots of leeway for the individual car - N-G does not apply to individual cars. Modeling the 1943 Delano picture that included only four box cars in Belan, New Mexico, all from eastern roads, and one a very rare road, is perfectly acceptable. If someone made a layout of that ATSF line, and 25% of the 100 XM's on the layout were from that line, then people would naturally say huh? Doesn't make sense (and highly improbable.) A banged up hopper being loaded - go for it - happened pretty often from the pictures I have seen (per-war and during WWII.)

So don't sweat it. Without calculating probabilities, what would the odds be that on a particular day, at that same coal mine, ONLY LV hoppers would be present, not a Southern or Interstate hopper in sight?

The odds of that, lacking any prior proof that a unit train of LV hoppers once visited the mine, becomes extremely unlikely, and most likely your modeling friends would raise a few eyebrows, at a minimum.

As we have noted, hoppers did not follow the N-G theory since most hoppers (but not all) were used in near-captive fleets, especially for soft coal. If the LV car was a boxcar, and only one arrived at the mine filled with equipment, then the LV car is fine. If the local food wholesaler received ten shipments, all in LV box cars, then there is a conflict - more raised eyebrows.

So for very small quantities of cars, N-G is irrelevant. When you are trying to model a fleet of boxcars, and perhaps flats and general purpose gons, and you want to create a fleet that is plausible, or even likely for the location and era you model, to include a large classification yard, then beware - it is possible for people to create a scene that while not impossible, is highly improbable, and that may also raise some eyebrows, with at least some modelers (for example, Dennis filling his yard with only 200 Tennessee Central box cars, and nothing else....) This is where N-G becomes a useful tool when collecting a fleet of XM's.

But in the end, I fully acknowledging that it is your railroad - do whatever floats your boat! It is up to you how many eyebrow raising looks you want from your closest friends ;-)

Dave Evans


---In STMFC@..., <gsc3@...> wrote :

I post to change no one's opinon, merely to hopefully entertain.    I model s.w. Virgina.  Ton of Southern and Interstate hoppers.  Now to model a single day in reality and follow photos, regardless of the G-N distribution model, there is one day when I have to model an empty Lehigh Valley hopper at a s.w. Virginia mine.  Thus there is an inherent philosophical conflict between modeling reality on one day and distribution models.
  Should one only model undamaged cars because they represent the highest range of freight cars on any given day?  Or model a damaged car you see on a given day?  I watched too much football this weekend and am trying to distract myself.  Still if you follow a distribution model does that philosophically mean you must model an extended time period rather than a single or small time period?  Or if you model a specific time or a single day, then any average or norm must be out the window?

George Courtney


Re: The G-N Distribution Model

George Courtney
 

I post to change no one's opinon, merely to hopefully entertain.    I model s.w. Virgina.  Ton of Southern and Interstate hoppers.  Now to model a single day in reality and follow photos, regardless of the G-N distribution model, there is one day when I have to model an empty Lehigh Valley hopper at a s.w. Virginia mine.  Thus there is an inherent philosophical conflict between modeling reality on one day and distribution models.
  Should one only model undamaged cars because they represent the highest range of freight cars on any given day?  Or model a damaged car you see on a given day?  I watched too much football this weekend and am trying to distract myself.  Still if you follow a distribution model does that philosophically mean you must model an extended time period rather than a single or small time period?  Or if you model a specific time or a single day, then any average or norm must be out the window?

George Courtney


Rail shipments of newsprint from Canada to the US, 1947

Dave Nelson
 

Posted because the topic has come up multiple times in the past.

 

https://babel.hathitrust.org/cgi/pt?id=uiug.30112104062325;view=1up;seq=44

 

About 82% of all rail shipments of newsprint were from Canada to the US.  Canada provided over 80% of all US newsprint usage. 

 

Not everything shipped from Canada came by rail… 10-24% of the grand total of newsprint imports would move by ship, which, IIRC has not been previously mentioned here.

 

Table 19 cites done delivered to individual states.  The average carload tonnage for 1947 (determined by the 1% sample sent to the ICC) was 29 tons / carload and so if you are numerically inclined you can easily convert the posted tonnage to estimate rail carloads.

 

Last, if any of you have the annual car load commodity classification reports do note that all ship to rail movement would be classified as received from another carrier and not received locally (as if the ship was another class 1 railroad).

 

Dave Nelson


Re: The G-N Distribution Model

SUVCWORR@...
 

And you can carry the reality even further by having off staging storage for those cars that might only show up once a year.

For example, the Elkland Leather tank car carrying spent tanic acid may only move to a processing facility once every 3 or 4 months in reality.  So the car when returned empty is taken off the layout and stored for the next Y number of operating sessions.  You determine the Y.  In the case of this car and operating twice a month it would sit out for 6 - 8 sessions.  All it requires is another car card that indicates when it is to be cycled onto the layout.

I use Elkland leather as my example because I am currently seeking information on these cars to build one of their tank cars which will appear perhaps only once every 6 months.  The planned fleet for the Western Pennsylvania Model Railroad Museum has a number of these cars which all are cycled following the same method.

Rich Orr


-----Original Message-----
From: destorzek@... [STMFC]
To: STMFC
Sent: Tue, Sep 5, 2017 6:57 pm
Subject: RE: [STMFC] The G-N Distribution Model



I guess I am really more attuned to operations rather than modeling a static collection. Where I see value in the G-N distribution model is in building the layout FLEET. That is the global set that all trains will be built from.  

One of the layouts I operate on uses the time tested CC&WB (Car Card and Way Bill) system with four cycle waybills. The owner has chosen to include the MTy moves as part of the cycle, and that makes what Dave mentions entirely possible; MTy hopper trains are mostly home road cars with a few captured foreign cars thrown in, WB hauler trains have mostly western road MTys, while EB hauler trains are just the opposite. Routing of particular road's MTys are repetitious, because that road's cars always go home through the same gateway. However, the consist of any particular train, and by extension the yard that's building it, is entirely random, as it should be. ALL are within the realm of plausibility, because the fleet of cars they are drawn from match a reasonable distribution of the national fleet. 

Dennis Storzek



Re: The G-N Distribution Model

destorzek@...
 

I guess I am really more attuned to operations rather than modeling a static collection. Where I see value in the G-N distribution model is in building the layout FLEET. That is the global set that all trains will be built from.  

One of the layouts I operate on uses the time tested CC&WB (Car Card and Way Bill) system with four cycle waybills. The owner has chosen to include the MTy moves as part of the cycle, and that makes what Dave mentions entirely possible; MTy hopper trains are mostly home road cars with a few captured foreign cars thrown in, WB hauler trains have mostly western road MTys, while EB hauler trains are just the opposite. Routing of particular road's MTys are repetitious, because that road's cars always go home through the same gateway. However, the consist of any particular train, and by extension the yard that's building it, is entirely random, as it should be. ALL are within the realm of plausibility, because the fleet of cars they are drawn from match a reasonable distribution of the national fleet. 

Dennis Storzek


Re: The G-N Distribution Model

devansprr
 

Tim,

Not quite. How you define equally likely is important.

Is it equally likely that Dennis' yard contains all 200 TC XM's, and 200 SPECIFICLY numbered PRR X29's - yes. The nastiness of the statistics in this situation, properly applied, is that the probability of any 200 PRR XM's filling Dennis' yard is orders of magnitude greater than all 200 TC XM's, unless it was TC XM delivery day (assuming they were all delivered on the same day, also very unlikely.)

So the perception of randomness (how many XM's from each road) is actually driven by very different probabilities. And the math is very nasty - I only know that is way beyond my skill set, and if I recall from back in 2009, I am not sure statisticians can provide a probability for each "randomness" without throwing massive amounts of computational power to the problem (there are too many variables) - I do not know if the mathematical tools even exist for such a problem. And no one cares about the individual probabilities.

Personally I only hope to present a fleet that is likely for the era, based on the distribution of wood vs. steel, 36' vs 40' vs 50', and reporting marks/heralds. Adding a TC XM to my fleet is not a priority for me. A reasonable sampling of PRR, NYCentral and B&O steel XMs is.

Dave Evans

(For your coin toss analogy, the odds of tossing 50 heads in a row is one probability. The odds of throwing 50 heads in a row out of 100 coin tosses, that totaled 50 heads and 50 tails, is much less likely - i.e. tossing 50 heads in a row, along with two streaks of tails in a row that totaled 50.)


Re: The G-N Distribution Model

devansprr
 

Schuyler, Claus,

Dennis's response is educational, and correct. If it is the same day every day, you can vary from N-G by huge amounts and stay well within the acceptable realm of statistics* - application of the N-G theory really depends on you modeling objectives.

Personally I am more into operations than fine scale modeling (so I am a bit of a lurker on this group), but from an operational standpoint, if I can ever afford to model small portions of the PRR main between Harrisburg and Johnstown, PA, I would think it pretty cool if an engineer departing west bound out of Harrisburg staging with a train consist of almost exclusively western RR MTs (100 car WB MTY trains happened about ten times a day during WWII on the PRR's middle division, nearly all of them XM's), noticed the lack of any NYCentral box cars in his consist, especially when he sees a bunch of them in the EB freights passing him by.

My response would be that the PRR sent any NYcentral MT's north out of Harrisburg, not west (by the way, there wouldn't be any Southern MT's in that train either, nor RDGs and several other east coast roads). And if I am fortunate enough to model a compressed version of the Altoona yards, I would want to make sure that any NYCentral XM's unloaded in Altoona, are placed by the yard crew into the one out of seven EB freights departing Altoona that headed up the Bald Eagle branch at Tyrone, rather than add to the massive flow of freight towards Harrisburg. And by the way, any newly MT'd RDG XM's that could not find a load in Altoona, would be added to a specific EB local freight out of Altoona destined for interchange with the RDG in Harrisburg (nearly 120 miles away.)

We each have our interests. Mine is to create a representative WWII fleet, and then use it to create prototypical routing scenarios for the operating crews. Both have some rewarding challenges.

Dave Evans

* Statistically speaking, coming up with the probability of a specific set of cars being on your layout is an astronomically intensive mathematical calculation. Assuming Dennis' layout does not interchange with the TC, having 10 TC XM's in his yard is probably akin to only ever playing poker once, and being dealt a royal straight flush on that one hand. Whereas seeing twice the expected number of PRR XM's may not be any more improbable than being dealt a pair... The analogy with cards holds well. Trying to calculate the probability of getting specific cards playing blackjack at a Casino after seven decks have just been shuffled is a significant calculation, whereas calculating the odds of getting black jack, with any number of different card combinations, is much easier. During WWII, with at least 210 North American RR's fielding XM's, the odds of a specific distribution across 200 cars are pretty astronomical. That is why I settled on the proposed distribution in my last message - it would deliver one of the highest probability consist totals over 5 operating sessions. Would it be "correct" - no - no distribution would be most "correct". But if instead every car in the 200 XM's departing staging were from a different railroad, now the odds are becoming pretty astronomical that it would ever happen on the prototype. Not impossible, but very improbable. The prototype police will notice that.


Re: The G-N Distribution Model

Tony Thompson
 

Tim wrote:

 

It is only over time, with MANY samples, that the G-N distribution is perceived
to be true. So from a modeling perspective, if you choose to model ONE DAY in
the life of a railroad, it literally doesn't matter what cars you choose to use.
The fact that those twelve PRR cars only occur once in 10,000 days* should not be
an impediment to modeling at all. You just happened to choose THAT day. :-)


         Certainly true, Tim. But do note that the number of data sets that Tim and Dave examined was pretty far short of 10,000. It is interesting to me that the G-N idea can be observed in a limited number of samples.
         Obviously what many of us would like to achieve is a REASONABLY typical set of freight cars on the layout. For my own operating sessions, there has never been the same set on the layout twice. But I do look to see if I have at least one car from PRR, NYC and B&O, and am aware of other major car fleets like MILW and ATSF. No, I don't do statistical calculation nor even yogurt calculation. But I want to try and reflect what I believe should be reasonably probable.
          Speaking just for myself, I would not choose any really off-center grouping like 12 PRR cars. Sure, it could happen. That's not the point, in my view. I could also have one car each from MNS, IN and West India Fruit. Likely? not exactly. Possible? Sure, but I'd say, "so what?" As in so many things in model railroading, I prefer to try and represent the typical.

Tony Thompson             Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705         www.signaturepress.com
(510) 540-6538; e-mail, tony@...
Publishers of books on railroad history






Re: The G-N Distribution Model

anthony wagner
 

Don't forget B&O's thousands of box cars. Tony Wagner


On Tuesday, September 5, 2017 2:23 PM, "'Aley, Jeff A' Jeff.A.Aley@... [STMFC]" wrote:


 
I’ll save you the trouble of looking up the yogurt:
 
[QUOTE]
I have countered numerous times that 1) one train does not make a sample and 2) the distribution hypothesis has nothing whatsoever to say about the composition of a single train. I will fall back, once again, on my Yoplait Yogurt in the shopping car example: I do not purchase Yoplait at regular prices (tho my daughter may slip in one or two when I'm not looking); However when they are half off I [buy] a whole lot of them. The composition of my carts therefore varies considerably when examined individually but the average number of yoplaits I buy over numerous trips is probably pretty constant over time if enough trips are recorded and analyzed.
[END QUOTE]
 
I presume that in the 6 years since Dave posted the above, he has consumed a large quantity of yogurt.
 
Regards,
 
-Jeff
 
From: STMFC@... [mailto:STMFC@...]
Sent: Tuesday, September 05, 2017 10:03 AM
To: STMFC@...
Subject: RE: [STMFC] The G-N Distribution Model
 
 
One of the problems people seem to have in accepting the theory is in the
failure to understand that average does not mean instance. I recommend a
search in the archives for the word yogurt (possibly Yoplait yoghurt),
There should be two hits, one from me and a reply from Tony. Please find
and read.

Basically a single train will never conform to the average. It takes 20-30
real freight trains to conform to the average. How many times do you move
1200-2000 cars across your model railroad? So let's get rea herel: FIRST,
not everybody on the list is a model railroader moving 12-18 car trains.
Some of us are historians. Me for instance. The theory is about real
railroads, not model railroads. SECOND, A few of us use simulators where
60-80 car consists moving across a 100 mile district are the norm. Me for
instance. Which is to say I can apply the theory quite directly because my
choice of expressing historical railroading has no compression in its
representation whereas physical model railroaders are faced with huge
compression requirements and so cannot do a dozen whole 60-80 car consists.

What physical model railroaders can do is make reference to the theory when
making purchases so they're not too overloaded with the arcane AND give some
consideration in the planning stages of how one might cycle cars on and off
their layout so as to achieve greater variability in the arcane because they
do show up.

Dave Nelson



Re: The G-N Distribution Model

Tim O'Connor
 


Dennis, you've hit on an important point I also tried to make. To properly
model prototypical freight car movements, you need a LOT of excess freight
cars. "Randomness" is what you are describing, not chaos. In a truly random
distribution, all possible outcomes (composition of freight cars in your yard
on a particular day) are EQUALLY LIKELY. (The number of possible outcomes is
astronomically large.)

It is only over time, with MANY samples, that the G-N distribution is perceived
to be true. So from a modeling perspective, if you choose to model ONE DAY in
the life of a railroad, it literally doesn't matter what cars you choose to use.
The fact that those twelve PRR cars only occur once in 10,000 days* should not be
an impediment to modeling at all. You just happened to choose THAT day. :-)

Tim O'

* swag



Dennis Storzek wrote

The G-N distribution statistical study is certainly worthwhile in the quest for absolute truth, but long ago I decided that I'd stick with Chaos Theory instead. Not really a true example, it is a catchy name, and suits well enough. It basically says you need a large sample, over a long period, to see the G-N distribution. Therefore, the cars visible on my layout THIS DAY are not a large enough sample, and the consist of just one train is even less so.

So, although the G-N distribution predicts there should be five PRR boxcars in my yard, and there happen to be twelve, it doesn't much bother me... if you want to see something closer to the predicted distribution, you'll just have to come back tomorrow... of course, it's never tomorrow on my layout.


Mountain States RPM

Blaine
 

Railroad prototype modeler meets have finally come to the Intermountain Area! This Saturday! 

Connect with friends and fellow prototype modelers at the only prototype modeling meet in the Intermountain West. Enrich your skills by interacting over detailed models and content-driven presentations. Nationally recognized historians and modelers. Clinicians from Washington, California and Utah: Rick Selby will present on Per Diem boxcars; Brian Rutherford will present 'Modeling from Prototype Inspiration: The WP and the Inside Gateway.' Kevin Packard will share tips and tricks for photorealistic weathering with a live presentation. Jere Ingram will present an introduction to modular signal systems. Blaine Hadfield will give an executive overview of ExactRail from its origins to the present. 

Railfan the historic trans-con from the machine shop and roundhouse. Shop at Spring Creek Trains. Meet with Canon & Company's David Hussey and see his amazing line of parts and boxcar kits. The Union Pacific Historical Society will be in attendance, and Scale Trains will have a display table of future releases. Enjoy our barbecue on the patio. Raffle Prizes from Intermountain, Kadee, ScaleTrains, ExactRail, and Tangent. Fellowship. 

Bring your models! There is no cost to display. There are no contests. 

Mountain States RPM is at the historic Evanston Roundhouse & Machine Shop (1500 Main st.) in Evanston Wyoming. Doors open at 9:00 a.m. on Saturday, September 9th. The raffle begins at 5:15 p.m. See our Facebook page Mountain States RPM for more information. 

Admission is $20.00. Kids under 13 are free with paid adult admission. 

We can't wait to see you there! 

Blaine Hadfield, 
Mountain States RPM



Re: The G-N Distribution Model

destorzek@...
 




---In STMFC@..., <Jeff.A.Aley@...> wrote :


[QUOTE]

I have countered numerous times that 1) one train does not make a sample and 2) the distribution hypothesis has nothing whatsoever to say about the composition of a single train. I will fall back, once again, on my Yoplait Yogurt in the shopping car example: I do not purchase Yoplait at regular prices (tho my daughter may slip in one or two when I'm not looking); However when they are half off I [buy] a whole lot of them. The composition of my carts therefore varies considerably when examined individually but the average number of yoplaits I buy over numerous trips is probably pretty constant over time if enough trips are recorded and analyzed.

[END QUOTE]

============================

You mean the whole Yoplait thing is just about yogurt? I must have missed the original discussion six years ago, but when I saw that phrase, I was sure the point was how the much studied FREQUENCY OF LETTERS of English language texts predicts a "consist" far different from that trade name. After all, the most common NP boxcar, no, wait, I mean the letter E, is entirely missing, yet the phrase  has two Ys and a P, letters that are hardly common. The "consist" of that trade name is just entirely wrong.


The G-N distribution statistical study is certainly worthwhile in the quest for absolute truth, but long ago I decided that I'd stick with Chaos Theory instead. Not really a true example, it is a catchy name, and suits well enough. It basically says you need a large sample, over a long period, to see the G-N distribution. Therefore, the cars visible on my layout THIS DAY are not a large enough sample, and the consist of just one train is even less so. 


So, although the G-N distribution predicts there should be five PRR boxcars in my yard, and there happen to be twelve, it doesn't much bother me... if you want to see something closer to the predicted distribution, you'll just have to come back tomorrow... of course, it's never tomorrow on my layout.


There are those people who take the G-N distribution even farther, breaking down the composition of each road's fleet into classes, and stating that the road's representation on the layout should match. That's all well and good, if that's what they want to do. For those of us who don't build resin (and I haven't built resin for years, I'm very sensitive to CA and got tired of walking around with perpetual sniffles), Chaos Theory offers an out. No doubt those five PRR cars predicted to be in the yard should be from four different classes, but if they happen to be five nice injection molded X29s, I don't let it bother me... that's just the way it is, on this particular day. Tomorrow will be different, but, of course, tomorrow will never come.


Dennis Storzek


Re: dry transfer lettring

thmsdmpsy
 

I'm still working on it, maybe a reliable printer soon.  Tom Dempsey



From: "Tim O'Connor timboconnor@... [STMFC]"
To: STMFC@...
Sent: Tuesday, September 5, 2017 10:57 AM
Subject: Re: [STMFC] dry transfer lettring

 
Lester

That set has long been on my "I wish I'd bought a couple more sets" list!
The 9601-11 and 9601-21 are excellent for making up freight car lettering.
And Dry Transfers are the way to go for composing whole words or large numbers.

Tim O'Connor



Review of the Clover House website shows the old alphabet Railroad Roman Combined condensed bold -white dry transfer lettering set, number 9600-11, which had 1/32", 3/64",1/16" and 3/32" letters and numbers is no longer available.  Excellent to make up car end lettering decals and changing numbers.  Has  anyone found another manufacture for a source for this style lettering?

Lester Breuer



Re: The G-N Distribution Model

Aley, Jeff A
 

I’ll save you the trouble of looking up the yogurt:

 

[QUOTE]

I have countered numerous times that 1) one train does not make a sample and 2) the distribution hypothesis has nothing whatsoever to say about the composition of a single train. I will fall back, once again, on my Yoplait Yogurt in the shopping car example: I do not purchase Yoplait at regular prices (tho my daughter may slip in one or two when I'm not looking); However when they are half off I [buy] a whole lot of them. The composition of my carts therefore varies considerably when examined individually but the average number of yoplaits I buy over numerous trips is probably pretty constant over time if enough trips are recorded and analyzed.

[END QUOTE]

 

I presume that in the 6 years since Dave posted the above, he has consumed a large quantity of yogurt.

 

Regards,

 

-Jeff

 

From: STMFC@... [mailto:STMFC@...]
Sent: Tuesday, September 05, 2017 10:03 AM
To: STMFC@...
Subject: RE: [STMFC] The G-N Distribution Model

 

 

One of the problems people seem to have in accepting the theory is in the
failure to understand that average does not mean instance. I recommend a
search in the archives for the word yogurt (possibly Yoplait yoghurt),
There should be two hits, one from me and a reply from Tony. Please find
and read.

Basically a single train will never conform to the average. It takes 20-30
real freight trains to conform to the average. How many times do you move
1200-2000 cars across your model railroad? So let's get rea herel: FIRST,
not everybody on the list is a model railroader moving 12-18 car trains.
Some of us are historians. Me for instance. The theory is about real
railroads, not model railroads. SECOND, A few of us use simulators where
60-80 car consists moving across a 100 mile district are the norm. Me for
instance. Which is to say I can apply the theory quite directly because my
choice of expressing historical railroading has no compression in its
representation whereas physical model railroaders are faced with huge
compression requirements and so cannot do a dozen whole 60-80 car consists.

What physical model railroaders can do is make reference to the theory when
making purchases so they're not too overloaded with the arcane AND give some
consideration in the planning stages of how one might cycle cars on and off
their layout so as to achieve greater variability in the arcane because they
do show up.

Dave Nelson


Re: F&C's Pennsylvania F28 Well Hole Flat Car kit #8450 question

Gatwood, Elden J SAD
 

Bill;

Also, IIRC, the HO trucks (2F-F2 and later 2F-F3 (both "Buckeye", but different load limits) were bagged under the Crown Custom Products label, as well.

Elden Gatwood

-----Original Message-----
From: STMFC@... [mailto:STMFC@...]
Sent: Tuesday, September 05, 2017 2:11 PM
To: STMFC@...
Subject: [EXTERNAL] Re: [STMFC] F&C's Pennsylvania F28 Well Hole Flat Car kit #8450 question



Bill,

Railworks has been out of business for many years, but these cars and trucks show up on eBay and at the PRRT&HS annual meeting with some frequency.

Regards

Bruce




Bruce F. Smith

Auburn, AL

Blockedhttps://www5.vetmed.auburn.edu/~smithbf/

"Some days you are the bug, some days you are the windshield."




On Sep 5, 2017, at 11:03 AM, fgexbill@... <mailto:fgexbill@...> [STMFC] <STMFC@... <mailto:STMFC@...> > wrote:



Based on Rich's tabulation it seems like the F33 would be a better choice. I just did search for Railworks and did not find anything relevant. Are the brass trucks from Railwoarks?

Bill Welch


Re: [EXTERNAL] Re: nitpicking PRR x26 boxcar Youngstown doors in HO Scale

Gatwood, Elden J SAD
 

Guys;

There was a lot of discussion about this before the RP CYC article came out, and yes, there were a LOT of different doors on the X26 after the rebuilds: "new" Y-town doors, ex-X28 splices, Carbuilder splices, reinforced and rebuilt wooden doors, and some doors I have never see another example of. If there was a greater reason to say, "work from a photo", I cannot imagine.

Are we talking about THIS door?

http://digital.hagley.org/PRR_09897?solr_nav%5Bid%5D=58a30c6e456023b4f893&solr_nav%5Bpage%5D=0&solr_nav%5Boffset%5D=0

Anyway, I will be doing more digging on this as the PRR box car thing unfolds, and I will let you know what I find.

Elden Gatwood

-----Original Message-----
From: STMFC@... [mailto:STMFC@...]
Sent: Tuesday, September 05, 2017 12:38 PM
To: STMFC@...
Subject: [EXTERNAL] Re: [STMFC] nitpicking PRR x26 boxcar Youngstown doors in HO Scale



I hate to be contrarian, and Ben has corrected me before, but the X26 photos I have available show most commonly the 5/4/5 door as on the USRA car in Bryian's photo.

38333, 45466, 46096, 46511, 540176, 564311, 565081 all have 5/4/5 doors without evidence of welded halves.

I also have 45927 with a 7 panel Superior door, and 45939 with a 3 panel door like the X29 with well defined welded halves.

I don't have any X26 photos with any other type of Youngstown door, although I do not doubt anyone's reporting of such doors.

Steve Hoxie
Pensacola FL

42201 - 42220 of 194687