Daily advance rate comparisons

Discussions on every day life in the Weimar Republic, pre-anschluss Austria, Third Reich and the occupied territories. Hosted by Vikki.
Post Reply
User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#91

Post by Qvist » 25 Jan 2007, 11:09

Hi Jon

I don't think they both can be (advance rate and duration) - they would either, as you say, double-count (if you score high duration low) or cancel each other out (if you score high duration high).

The only solution I can think of is to take duration out of the model, and then stick to comparing operations of (roughly) similar duration.

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#92

Post by JonS » 25 Jan 2007, 11:20

Perhaps, but then for consistency you need to take out DISTANCE for exactly the same reasoning.
Last edited by JonS on 25 Jan 2007, 11:21, edited 1 time in total.


User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#93

Post by Qvist » 25 Jan 2007, 11:21

Why?

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#94

Post by JonS » 25 Jan 2007, 11:23

RATE = DISTANCE / DURATION

If you intend to take out DURATION because it is already accounted for in RATE, then you also need to take out DISTANCE because it, too, is already accounted for in RATE.

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#95

Post by Qvist » 25 Jan 2007, 11:42

Ah, right, I see. Sorry, I meant just switch to using duration as the static factor rather than distance - hence you won't need to score it, as it is similar in the cases compared. And distance will then also autmatically exoress the advance rate, just like duration does if you use distance as a static factor.

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#96

Post by JonS » 25 Jan 2007, 22:09

Ah, ok. I've always been thinking in terms of a general model where any - or none - of the factors can be controlled for.

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#97

Post by Qvist » 26 Jan 2007, 00:20

That'd be ideal of course, but this solution would perhaps remove some problems and insecurities, at least as a start.

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#98

Post by JonS » 26 Jan 2007, 01:38

:) I view it the other way ...

I think it might be time to restate the model:

1) DISTANCE (long/med/short) - rough measure of distance covered during advance.

2) DURATION (long/med/short) – rough measure of duration of operation

3) RATE (fast/med/slow) – rough measure of daily rate of advance

4) MASS (large/med/small) – rough measure of size of attacking force (using eg divisions, corps or armies as initial unit of measure)

5) ENEMY (strong/med/weak) – rough measure of effectiveness of opposition to offensive.

6) INFRASTRUCTURE (poor/med/good) – rough measure of pre-existing native, or civilian, infrastructure in area of operation usable for logistics purposes

7) ENVIRONMENT (awful/med/benign) – rough measure of difficulties imposed by the area in which operation occurs.

I've included all three of RATE, DURATION, and DISTANCE because of the previous discussion. You can easily control for one of the three (which n essence means it drops out of the comparison) but in doing so the other two become significant, so I don't think that leaving one of the three out of the model altogether - notably DURATION per the previous discussion - is a useful approach.

All factors are only measured in relation to other operations under consideration, not in any absolute sense. So, depending on the particular set of ops being looked at, the ‘score’ for particular factors in a given op can change (eg, DISTANCE for Op Bluecoat is ‘short’ or 'med' compared to Op Market Garden, but ‘long’ compared to Op Windsor).

Once all ops have been given a rating for each factor, convert factors to numeric ‘scores’, 4 for the first listed rating (long, fast, strong, poor, awful), 2 for ‘med’, and 1 for the last listed rating (short, slow, small, weak, good, benign). Total the ‘score’ for each op, then compare and analyse/conclude.

Eg (All that follows is just an example, based on my half-arsed understanding of the two ops. Feel free to quibble with the ratings, but bear in mind that accuracy isn’t terribly important in terms of the working of the model*)

[begin example]
Step 1: rating

Op Barbarossa, AGN:
1) DISTANCE - long
2) DURATION - long
3) RATE – med
4) MASS - small
5) ENEMY – weak
6) INFRASTRUCTURE – med
7) ENVIRONMENT – benign

Op Bagration:
1) DISTANCE - med
2) DURATION - long
3) RATE – med
4) MASS - large
5) ENEMY – med
6) INFRASTRUCTURE – med
7) ENVIRONMENT – med

Step 2: scoring

Op Barbarossa, AGN:
1) DISTANCE - 4
2) DURATION - 4
3) RATE – 2
4) MASS - 1
5) ENEMY – 1
6) INFRASTRUCTURE – 2
7) ENVIRONMENT – 1
Total = 15

Op Bagration:
1) DISTANCE - 2
2) DURATION - 4
3) RATE – 2
4) MASS - 4
5) ENEMY – 2
6) INFRASTRUCTURE – 2
7) ENVIRONMENT – 2
Total = 18

Step 3: Conclusion
Bagration was a little more difficult, logistically, than Barbarossa/AGN, and implies that the Soviet logistics system in mid ‘44 was a little more mature/sophisticated/capable than the German logistics system in mid 41’. The difference was primarily due to the greater scale of Bagration, and to a greater level of German resistance, at least in the initial stages of the op.
[end example]

Setting steps 1 and 2 up as a table, with factors along the top and operations down the side, would greatly assist in intra and inter comparisions.

Jon

* although better accuracy would help in terms of validating the model and highlighting unexpected results … although that doesn’t necessarily mean the model is wrong. It could mean that factors one hadn’t previously considered, or had discounted, actually have a significant influence.

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#99

Post by Qvist » 26 Jan 2007, 10:21

Yes, I know this is the sort of model you had in mind, but the problems with the logic of duration and rate are still the same. As presently consituted, they cancel out in cases where the distance rating is similar - you get exactly the same net rating for a slow advance as for a speedy advance over the same distance. And that is an irrational result, AFAICS.

Will return with more detailed comments when I have the possibility. And have had a bit of a think.

cheers

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#100

Post by Qvist » 29 Jan 2007, 12:42

* although better accuracy would help in terms of validating the model and highlighting unexpected results … although that doesn’t necessarily mean the model is wrong. It could mean that factors one hadn’t previously considered, or had discounted, actually have a significant influence.
I agree. But this particular point concerns the fundamental logic of it, and that has to be right. Put differently, the model needs to function in a way that does not produce results that appear intuitively absurd, which presently it does.

I think the problem is in the nature of the relation between duration, distance and advance rate.

Each of these are implicit in the relation of the two others. That alone means that all three cannot possibly be included, because - and rein me in if I am mistaken here - it neccessarily means that any two operations where any one of the variables are similar will always get exactly the same score - one of the values being similar, the other two will in sum equal a medium rating. If you have a similar advance distance rating in two cases, say one that would be rated medium, then that advance can only be achieved at a fast pace over a short time, at a slow pace over a long time, or at medium pace over a medium amount of time - any other combination will not produce a similar advance distance.

Unlike distance and advance rate (where higher values are always more demanding logistically than lower values, AFAICS), the impact of duration depends on the total picture - it does not seem to relate to the others in a straightforward way. Itis logistically easier to sustain an operation over a given distance at a slow rate of advance than it is to do so at a fast rate of advance (otherwise, why do logistical problems typically result in a slower rate of advance? Right now, the model is assuming f.e. that the western allies could have alleviated their logistical problems by advancing to the German border much quicker than they did in the fall of 1944 - quite absurd).

On the other hand, it is more demanding to sustain a given rate of advance for 50 days than to do so for 5 days.

Hence, it seems to be true that in some cases it is intuitively correct to score long duration high, whereas in other cases it is intuitively correct to score long duration low.

If we do not want to use one of the three variables as a static criteria for comparison, we AFAICS have the following possibilities:

1. Include Duration and distance, leave out advance rate. High duration then needs to be scored low, because otherwise the model would consistently and neccessarily reward a low rate of advance. This would however also generate some irrational results - ie, a very long advance over a very sustained period would only score medium (high for distance, low for duration).

2. Include duration and advance rate, leave out distance. High duration then needs to be scored high, because otherwise we would consistently and neccessarily reward short advances. The relation between the two seems intuitively right - if both values are high, you also have a long advance, if both are low you nececssarily have a short one. However - it would assume that duration and advance rate are equal factors, and that it is equally demanding to sustain a long and slow advance, and a brief and fast one, which seems questionable unless the distance is roughly similar.

3. Include distance and advance rate, leave out duration. This seems to work too, but in a different way - it will score highly if both distance and advance rate is high, which has no systematic relation to duration, but is intuitively correct IMO. However,it will score any combination of quick and short just as highly as any combination of long and slow - much as in 2.

I would conclude that we can't have all three factors, and that we can't escape from restricting comparison to cases where whatever factor we omit from the model is roughly similar - otherwise the assumptions logically inherent in the model simply will not be tenable, AFAICS. Furthermore, that advance rate must be one of the two included, because if it is distance and duration, then there is no way in which duration can be scored that does not produce an irrational result in some cases.

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#101

Post by JonS » 29 Jan 2007, 22:33

Hi Qvist,
I haven't given up on this, but I've been too busy with other stuff to given it some concentrated thought. I'm going to carve up your previous post and respond quickly to it bit by bit, but hopefully I'll be able to do it more justice at a later date.
Qvist wrote:the model needs to function in a way that does not produce results that appear intuitively absurd, which presently it does.
I agree with the general jist of this, but not that it necessarily produces absurd results. Furthermore - and this is a very general point - intuition is a somewhat variable and unreliable baseline.
I think the problem is in the nature of the relation between duration, distance and advance rate.
Agreed, and I think our 'difference' here is in terms of degree, rather than principle.
That alone means that all three cannot possibly be included
I don't agree with that though.
because ... it neccessarily means that any two operations where any one of the variables are similar will always get exactly the same score - one of the values being similar, the other two will in sum equal a medium rating.
I started playing around with numbers for these three ratings the other day, and IIRC this isn't always the case, but I need to do more playing with numbers.

However, if you hold DURATION steady and send DISTANCE to the extremes (meaning RATE does too, but inversely), then it is true, but on the other hand that can seen as an improvement over 'just' using DISTANCE and DURATION, when a long slow advance was 'better' than a long fast advance :)
Right now, the model is assuming f.e. that the western allies could have alleviated their logistical problems by advancing to the German border much quicker than they did in the fall of 1944 - quite absurd).
Well as it stands, it isn't absurd. The advance from the Seine to Brussels (or thereabouts, and using equivalent points for the US) was short DURATION by any measure, long DISTANCE, and fast RATE. However, even this was only acheived for the short period that it was by various - well known - extreme measures. Expedients, if you like. Montgomery and Bradley did very well to get as far as they did, as fast as they did, but they only did it by 'breaking' their logisitics system, and they had to stop where and when they did because it was broken so badly.

So, IMHO, this particular advance validates the model. It was good while it lasted, but - agasin IMHO - it shouldn't score higher because of it's improvised nature, and becuase longer, farther advances would reflect a more mature or sophisticated log system, rather than one that was made to work by grounding half the available forces, and those advances are the ones we should laud, from a log POV.
Hence, it seems to be true that in some cases it is intuitively correct to score long duration high, whereas in other cases it is intuitively correct to score long duration low.
This is a problem :)
1. Include Duration and distance, leave out advance rate.
I think we agree this isn't valid?

I need to think about the other two.
I would conclude that we can't have all three factors, and that we can't escape from restricting comparison to cases where whatever factor we omit from the model is roughly similar
I don't necessarily agree with the first bit, and the second bit makes the model less than useful.

As I was walking to work this morning I was playing around with using constants and multipliers to enhance the effect of any single factor (eg: DURATION + DISTANCE + 2 x RATE, or DURATION x DISTANCE x RATE, or some approach) ratehr than straight addition. It would woork, but I'm a bit uncomfortable about that approach since it means somewhat arbitrarily elevanting the importance of one - or a few - factors above others, which is something I was trying to get away from in the first place (ie, originally when RATE was the only thing that mattered). Now, it is probably true that one or a few of the factors are more important than the others, but I don't really feel qualified to assess which ones and by how much. It can also very easily lead into some very tricky ground in which various factions vie for enhanced mana for 'their' factor.

Regards
Jon

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#102

Post by Qvist » 30 Jan 2007, 13:27

Hi Jon
I agree with the general jist of this, but not that it necessarily produces absurd results. Furthermore - and this is a very general point - intuition is a somewhat variable and unreliable baseline.
I know, I agree that is important to remember, but also this cannot be ignored, as we have no other methodology for checking the assumptions of the model - and their validity can't be taken for granted, after all.
Quote:
That alone means that all three cannot possibly be included

I don't agree with that though.
Quote:
because ... it neccessarily means that any two operations where any one of the variables are similar will always get exactly the same score - one of the values being similar, the other two will in sum equal a medium rating.

I started playing around with numbers for these three ratings the other day, and IIRC this isn't always the case, but I need to do more playing with numbers.
You seem to be right - what I failed to consider was that Duration, distance and Advance rates are not internally linked, but are all scored from high to low separately, so it depends entirely on where you put the dividing points on the scale of each.

Played with some numbers too:

DUR DIS AR Score pts avg
10 50 8 LLL 1
10 160 16 LMM 1,7
10 250 25 LHH 2,3
25 50 2 MLL 1,3
25 160 6,4 MML 1,7
25 250 10 MHM 2,3
50 50 1 HLL 1,7
50 160 3,2 HML 2
50 250 5 HHL 2,3

Ranked by pts avg:

DUR DIS AR Score pts avg
25 250 10 MHM 2,3
10 250 25 LHH 2,3
50 250 5 HHL 2,3
50 160 3,2 HML 2
10 160 16 LMM 1,7
25 160 6,4 MML 1,7
50 50 1 HLL 1,7
25 50 2 MLL 1,3
10 50 8 LLL 1

Here, the following scale has been used:

Duration: 10=low, 25=medium, 50=high
Distance: 50=low, 160=medium, 250=high
Advance rate: less than 10=low, 10-20=medium, 20+=high

This seems to point to some things:

- With just three levels of scoring, it matters enormously just where you put the dividing line between each score, for which there are no very obvious criteria.

- It seems the model consistently score longer total distances more highly than lower total distances, regardless of duration and AR. This would be the previously noted effect of AR/Duration cancelling out relative to a given distance, although they do not, contrary to what I thought, neccessarily produce similar results, they just tend to. This means that effectively, the model treats advance distance as much more important than advance rate or duration. Not sure if this is good or bad, probably it is bad.

- Generally, it also tends to treat duration and AR as interchangeable entities - low AR/high duration and high AR/low duration in all cases but one produces a similar result for the same overall distance.

One result I find problematic:

50 50 1 HLL 1,7
25 50 2 MLL 1,3
10 50 8 LLL 1

A short overall distance achieved at a real snail's pace over a long period of time score much higher than a similar advance in a much shorter time.

But overall, these are more "intuitively correct" results than I had expected, I must say. Unless I am missing something. Which has been known to happen.
Quote:
Right now, the model is assuming f.e. that the western allies could have alleviated their logistical problems by advancing to the German border much quicker than they did in the fall of 1944 - quite absurd).

Well as it stands, it isn't absurd. The advance from the Seine to Brussels (or thereabouts, and using equivalent points for the US) was short DURATION by any measure, long DISTANCE, and fast RATE. However, even this was only acheived for the short period that it was by various - well known - extreme measures. Expedients, if you like. Montgomery and Bradley did very well to get as far as they did, as fast as they did, but they only did it by 'breaking' their logisitics system, and they had to stop where and when they did because it was broken so badly.

So, IMHO, this particular advance validates the model. It was good while it lasted, but - agasin IMHO - it shouldn't score higher because of it's improvised nature, and becuase longer, farther advances would reflect a more mature or sophisticated log system, rather than one that was made to work by grounding half the available forces, and those advances are the ones we should laud, from a log POV.
Ah, but that can also be regarded differently, though I acknowledge your logic here too, as far as it goes (at least mostly, more on that later). You may be correct that the advance to Brussels does not reflect laudable logistical choices or a mature logistical system, but the model AFAICS is not designed to measure how laudable the logistical choices are - rather it is designed to measure how logistically difficult a given advance is, on the assumption that the greater the difficulty relative to the distance, the better the net logistical output - that is why we score high force density, difficult environment and bad infrastructure highly, yes? And as you point out, the quick and far advance to Brussels could only be achieved by recourse to extraordinary expedients, meaning that it was logistically difficult to achieve. If it had been slower, it would also have been simpler to supply.

As presently constituted, the model can't account for things like slowing down some advances in order to support others, at least not on an individual case basis. Also - it does not account for the price you may have to occasionally pay for achieving very good advances. I suppose the same point could be applied to Barbarossa, with the lengthy pause on the central sector in July/August, significantly due to the need to let logistics catch up. Perhaps we need to reflect that somehow, weighing the desirability of quick advance against the inevitable logistical problems that this creates. Not that I know how to, except that it probably is something that needs to be introduced as part of the context in evaluating the end result rather than as a part of the model itself. But that brings a whole new set of problems.
Quote:
1. Include Duration and distance, leave out advance rate.

I think we agree this isn't valid?

I need to think about the other two.
Yes, this seems to be the most defective combination.
Quote:
I would conclude that we can't have all three factors, and that we can't escape from restricting comparison to cases where whatever factor we omit from the model is roughly similar

I don't necessarily agree with the first bit, and the second bit makes the model less than useful.
Yes, the first seems to be less true than I thought (though still a problem). As for the second bit, true, but I would rather sacrifice scale of applicability than reliability.
As I was walking to work this morning I was playing around with using constants and multipliers to enhance the effect of any single factor (eg: DURATION + DISTANCE + 2 x RATE, or DURATION x DISTANCE x RATE, or some approach) ratehr than straight addition. It would woork, but I'm a bit uncomfortable about that approach since it means somewhat arbitrarily elevanting the importance of one - or a few - factors above others, which is something I was trying to get away from in the first place (ie, originally when RATE was the only thing that mattered). Now, it is probably true that one or a few of the factors are more important than the others, but I don't really feel qualified to assess which ones and by how much.
I've thought about this too - something more akin to combat efficiency calculations, where the factors take the form of modifiers to, say, advance distance. The problem as you note is how to set the factors, normally you need to do that by validating the model against a large set of well-researched historical cases, otherwise it would risk being merely speculative. But of course, assigning an equal value to all of them is strictly speaking no less speculative. :)
It can also very easily lead into some very tricky ground in which various factions vie for enhanced mana for 'their' factor.
Very true. But at least, this is merely a difficulty rather than an essential obstacle. :)

cheers

JonS
Member
Posts: 3935
Joined: 23 Jul 2004, 02:39
Location: New Zealand
Contact:

#103

Post by JonS » 31 Jan 2007, 00:12

Qvist wrote:I know, I agree that is important to remember, but also this cannot be ignored, as we have no other methodology for checking the assumptions of the model - and their validity can't be taken for granted, after all.
Sure.
the following scale has been used:

Duration: 10=low, 25=medium, 50=high
Distance: 50=low, 160=medium, 250=high
Advance rate: less than 10=low, 10-20=medium, 20+=high

This seems to point to some things:

- With just three levels of scoring, it matters enormously just where you put the dividing line between each score, for which there are no very obvious criteria.
Yes and no. Remember the model is supposed to be comparative rather than absolute. This means that the model relies on good judgement to be used usefully, since it can't be very prescriptive.

Also, the value assigned will depend. In one comparison a particular advance DISTANCE might be "long", while in another it might be "med" or "short". For example, the drive to Brussels is "long" compared to GOODWOOD, but "short" (or perhaps "med) compared to El Alamein to Enfidaville.
One result I find problematic:

50 50 1 HLL 1,7
25 50 2 MLL 1,3
10 50 8 LLL 1

A short overall distance achieved at a real snail's pace over a long period of time score much higher than a similar advance in a much shorter time.
I think you might be scoring in a different way than I intended. A short DISTANCE, long DURATION op (with a necesarily low RATE) would score 1 + 4 + 1 = 6, not (1+4+1)/3 = 2. Similarly a long, long, high op would score 4 + 4 + 4 = 12, not ( 4 + 4 + 4)/3=4

... although it occurs to me - after working that out - that it probably doesn't make a lot of difference ... :roll:

Oh, but a thought occurs. Given that the DISTANCE and DURATION interactions provide for nine (9) potential outcomes, perhaps there needs to be nine (9) potential scores for RATE.

*plays around a bit*

No, it's five (5) levels, not nine (9):
1) V.Low
2) Low
3) Med
4) High
5) V.High
See attached graph.

Numeric scoring is an issue ... is it 1, 1.3, 2, 2.5, 4, (maintaining the log scale, and sticking to the 4-as-max-score 'rule') or 0, 1, 2, 4, 8 (maintaining the log scale, but providing an 'incentive for long AND fast ops)?

Still, I think this might answer your concerns about the D-D-R interactions?
Ah, but that can also be regarded differently, though I acknowledge your logic here too, as far as it goes ... the model AFAICS is ... designed to measure how logistically difficult a given advance is, on the assumption that the greater the difficulty relative to the distance, the better the net logistical output - that is why we score high force density, difficult environment and bad infrastructure highly, yes?
Yes. See previous comments about gymnastics and degree of difficulty.
as you point out, the quick and far advance to Brussels could only be achieved by recourse to extraordinary expedients, meaning that it was logistically difficult to achieve. If it had been slower, it would also have been simpler to supply.
And scored lower as a result.
As presently constituted, the model can't account for things like slowing down some advances in order to support others, at least not on an individual case basis. Also - it does not account for the price you may have to occasionally pay for achieving very good advances. I suppose the same point could be applied to Barbarossa, with the lengthy pause on the central sector in July/August, significantly due to the need to let logistics catch up.
Yes, that is correct. And I agree with your suggestion that that should be external to the model, part of evaluating results. So the model provides a way of comparing diverse advances, then in the analysis phase you need to look at why and how different advances scored differently, and what the effects of acheiving particular scores were (eg, the mid-BARBAROSSA pause in Aug 1941, or the W.Allied halt on the German boder in Sept 1944)
I've thought about this too - something more akin to combat efficiency calculations, where the factors take the form of modifiers to, say, advance distance. The problem as you note is how to set the factors, normally you need to do that by validating the model against a large set of well-researched historical cases, otherwise it would risk being merely speculative.
Hmm. "Logistics efficiency calculations". Sounds like a PhD for the asking ... probably a little more ambitious than we can expect to thrash out here on the forum ;)
But of course, assigning an equal value to all of them is strictly speaking no less speculative. :)
Oh, I agree 100%, and I did wonder if you'd spot that gaping hole in my logic :P

Still, while setting everything to the same weighting (barring perhaps RATE per the above comments and following table) does introduce systemic speculation, it does avoid subjective speculation. Whether that is good, bad, or indifferent ... ;)

Regards
Jon
Attachments
DDR.JPG
DDR.JPG (8.36 KiB) Viewed 1073 times

User avatar
Qvist
Member
Posts: 7836
Joined: 11 Mar 2002, 17:59
Location: Europe

#104

Post by Qvist » 31 Jan 2007, 00:44

I think you might be scoring in a different way than I intended. A short DISTANCE, long DURATION op (with a necesarily low RATE) would score 1 + 4 + 1 = 6, not (1+4+1)/3 = 2. Similarly a long, long, high op would score 4 + 4 + 4 = 12, not ( 4 + 4 + 4)/3=4
Oh, yes - sorry, I did not give that much thought, and should of course have clarified it. As you will have figured out, I simply scored 1 for Low, 2 for medium and 3 for high.
Hmm. "Logistics efficiency calculations". Sounds like a PhD for the asking ... probably a little more ambitious than we can expect to thrash out here on the forum Wink
Yep, way beyond the horizon of the feasible, unfortunately. But one thing that did strike me is the possibility that someone, somewhere, may have done that job already. None spring to mind though.

As for the rest - will return after some thought.

cheers

Post Reply

Return to “Life in the Third Reich & Weimar Republic”