Fortress entries… and beyond

[robstitt]

Hello all,

I entered a random Fortress serial on reading Jagan’s 10-year announcement - congratulations to Ross, Jagan and to all have contributed to an amazing and constantly improving resource - and on following a series of links found a number of inconsistencies and errors. I don’t know enough about the database to know how representative these might be but thought it worth posting what I found for consideration.

Starting point: entered Fortress II serial FA698 in the yellow search box and selected Search

Clicked on the link ‘Fortress II FA698’ in the Aircraft Serials for Fa698[sic] - formatting issue

This brought up Fortress IIa FA698 but this was a Fortress II and if it had been correct should have been presented as IIA

I then clicked on the ‘Henk Welting’s Database’ link and found what I believe is a generic Fortress entry presented as Fortress GRIIa which raises a several issues

Then clicking on the ‘6’ beside the Fortress GRIIa there are a number of issues:

It’s not clear why there are six Fortress losses listed when there were several more - perhaps the extent of Henk’s research…?

FL545 should be FL454 - link goes to a Dakota

FH207 should be FK207 - link goes to a Hudson

The presentation of the marks is inconsistent throughout the table.

And regarding the table, it’s not clear to me how the Nos relate to the Aircraft Name. Clicking the ‘6680’ total against the Avenger for example doesn’t bring up Avengers.

The page intro refers to a proof of concept database and am wondering if the above should be accessible from the new database.

I’m likely missing an obvious explanation but hope these observations are useful.

Robert

[Franek Grabowski]

Robert
I think the reason is that the data sheet was created on a very raw data freely available. The more detailed sets are copyrighted, and thus it is not possible to use eg. AB books verbatim. On the other hand doing the job from the basics, ie. transcribing hundred of thousands of F.78s and other records would require extra hands and lot of time. It would have been great to have a cross check database of all aircraft but there must be people to do it and a well thought pattern prepared to enter the data.
Franek

[Andy Marden]

Robert,
I've been taking on a lot of the aircraft database work to take some of the pressure off Jagan. He is the "main man", so will be much better placed to answer some of your points.
As Franek has mentioned, the database was created by "scraping" a lot of data from other sources. In the case of FA698, whoever put in the incident on 26 Mar 43 (presumably Henk - RIP) entered the type as "Fortress IIa". This was then automatically carried forward to the CWGC entry.
It's very easy to amend, and I have just done that: https://www.rafcommands.com/database...php?uniq=FA698
In the case of FL545, someone has incorrectly entered the serial of the Fortress when entering the incident. I have now tidied the two entries:
https://www.rafcommands.com/database...php?uniq=FL454
https://www.rafcommands.com/database...php?uniq=FL545
In the case of the Fortress the very top line still reads "Fortress IIa" - Jagan will need to correct that.
FH207/FK207 is exactly the same, now corrected:
https://www.rafcommands.com/database...php?uniq=FK207
There is a thread for pointing out errors in the database here:
https://www.rafcommands.com/forum/sh...se-Corrections
I action comments as soon as I can. Thanks for pointing out the errors as you have done.
Regards,
Andy

[Franek Grabowski]

I think that creating a proper aircraft & airmen database is the task for the next 10 years at least. Mind you primary records do contain errors or inaccuracies, and those should be reflected to trace the source of error. That said, I can imagine such an ultimate database working magic, but would there be anyone interested in that, when it is ready?
Franek

[robstitt]

Thanks, Franek and Andy. I now have a better idea of how the database currently works.

I’ll follow the same path for each of the Fortresses I’m familiar with and pass on any suggested amendments using the error reporting thread.

Robert

[Franek Grabowski]

Robert
I think this should be discussed with Jagan. If you are willing to share complete or partial B-17 database, then I guess the system will pick up any errors.
Franek

[Jagan]

Hi Rob

Thanks for trying out the Databases. I will take the chance to talk a little bit more about the DBs and how they work. The key DBs in the site currently are

CWGC recorded Deaths
POW/Evader Database
RAF Serial Numbers Database
AIR81 Cas Files
The Losses DB - Either SEAC or Henks.

If you note - all the above five data tables have one common relation - the aircraft serial number - that one number ties different threads and presents it into one page. In that context the third database - i.e the RAF Serial Numbers Database is very important in this equation. That database - have over 350K serials listed - I believe it is 99% accurate as far as tying the number to aircraft name is concerned this list only ties the serial with the aircraft type. The sub-type/mark level information is only indicative and should not be taken as gospel.

in your case Fortresses https://www.rafcommands.com/database...Fortress&qt=FM 201 entries grouped by Serial prefix (or none).. if you want it in a listing format - https://www.rafcommands.com/database...Fortress&qt=FN . If the mark information is wrong etc, you just need to provide me a list of the serials and marks and I can have that updated.

An example is the Hurricane data - every entry has a Mark assigned to it - thanks to a DB update provided by member Geoff Sinclair. He submitted a bulk excel file that i could use to apply and thats the best way of doing it.
Another example are B-24 losses - Matt Poole had provided data that is now part of the DB and it shows up with his name against it.
Lastly John Englested's Hurricane losses are part of the DB as well.

Coming to the Losses DB - the section titled as "Henks Database" started off from a listing of dates and serials that henk used in his threads. and I believe he only tracked possible fatal ones . That makes up 2/3 of the data in there. Subsequently some data form Air Britain files have been added - I still call it 'Henks DB' to differentiate it from the SEAC list which I curate. Andy has been very gracious to keep this DB updated and he is slowly transforming it into a more reliable data set. But it no where claims to be comprehensive enough. Serial number errors will exist in this database till it is bought to Andy's notice and he will correct them. the errors creep in because theoriginal sources referred to by Henk and even some of the accident cards, AIR81 files had incorrectly recorded them in the first place.

To answer some specific bugs you reported

a generic Fortress entry presented as Fortress GRIIa
Yes its the generic "Fortress Category" but the software will pick the first Fortress types that appears in the list. -FIXED

It’s not clear why there are six Fortress losses listed when there were several more - perhaps the extent of Henk’s research…?
Yes it is limited to Henk's research.

And regarding the table, it’s not clear to me how the Nos relate to the Aircraft Name. Clicking the ‘6680’ total against the Avenger for example doesn’t bring up Avengers.
This is a bug, and the only line that it happens. Ideally the word "Uncategorized" should appear as this heading is a catch all. All other rows will provide results specific to the type. -FIXED


It would be great to add to these entries, or just remove/update all for a specific type. However to make the updates, it is easier if an excel or a google sheet formatted list is provided.
PS: You can submit images too - one per serial number and it gets added to the DB.

EDIT: Two of the bugs are now fixed.

[robstitt]

Many thanks, Jagan. Very useful and glad the input was of value.

Can you tell me which fields you would need for a Fortress update in Excel?

Would the file need to be copied to Andy?

And wondering how one would include a note regarding, for example, Fortress FL461 being taken back by the USAAF prior to delivery?

Thanks for yours too, Franek.

Robert

[Jagan]

HI Rob,

First Excel : Just three fields

Serial, Aircraft Type, Detailed Note (optional)

Sample Page
https://www.rafcommands.com/database...php?uniq=LE947
Sample Record
LE947, Hurricane IIc, "Block 10, Hawker Aircraft Ltd., RR Merlin XX , 24 and 27. 29 Sept 1943 - 24 May 1944"


Second Excel is for the Losses DB. This is a bit more complicated.

Date,
Aircraft Name , Serial
RANK1,NAME1,Number1 - Pilot (Can be split into three fields or not, I can manage)
RANK2,NAME2,Number2 - Co-Pilot
Sqn/Unit
Airfield Operated from
COUNTRY
Loss Details

its fine to just email me. I will do a mass import and then Andy gets to edit and add more detail if he wants to.

[GSinclair]

There is another Hurricane list coming hopefully this year with a number of corrections to the original list, plus details of conversions. Meantime no production Hurricane used the Merlin 24 or 27. The 27 was the planned engine for the Hurricane V, the batch of Merlin 27 made were converted to Merlin 25 for Mosquito. Packard Merlin 28 engines were only used in Hurricanes for a few test flights, Merlin XX were fitted instead and the 28 used in Lancasters. Also no trace so far of any official "Block number" for Hurricanes.