Information Storage Methods
Any large project requires that it be broken down into smaller parts. The methods I employed are outlined here and were continuously developed to assist with storage and distribution of the material acquired. They evolved over 25 years and are presented in hindsight as clear cut decisions. In reality, options were tried and discarded until methods proved worthwhile.
Strategy
Research can be broken down into three areas:
Perhaps the most important points to remember are:
Provides focus and clear targets
The challenge of the activity can significantly enhance the ego of the researcher. This can have severe lifestyle consequences, e.g. 'obsession'
For every part in the research activity there must be a simple strategy. This is to ensure that all material and knowledge gained can be dealt with quickly. No back tracking or 'stuffing around'. The long term strategy I devised for this research was:
The photographic scene appeared to be the most volatile, with over half the rolling stock disappearing with five years. It was crucial to act fast and not waste time on data research that eroded precious time
Ad-hoc over many years
Mainly sightings and notes. Used camera as 'recorder'. Historical aspect of photograph backgrounds not written out yet. Usually answered by "Why did I take that photo?". Once photography was finished, and data sources obtained, the work rotated to full time data recovery.
Photography
The basis for the photography was the N scale modeling interest. My modeling method was to apply scratch build bodies onto modified commercial underframes. After careful thought and watching others use randomly recorded detail photographs I evolved a photographic method that ensured sufficient photographs could be recorded with the minimum amount of time and effort. To establish this method two systems needed to be created:
The system for side and end identification needed to be simple enough to avoid crawling under a vehicle. Whilst many side and end identifiers were common, the application for photography seemed unique. I'm sure many people may feel this is an unnecessary step. Quite true. However there are many types of rolling stock that have no identifiers and there is value in a strategy that records the methods used.
For each type of different identification method a small 3"x5" card was filled out. The card contained
Cards are then used as reference. Vehicle groups associated with identification are clearly shown as 'Type xxx" where this would be 'Type PL1' from the example above.
As this is purely an arbitrary system, at some point the references to the descriptions must be given. This will assist in the conversion from your arbitrary system to other systems or the official naming convention used.
The number of photographs determined to be the absolute minimum were ten photos:
This method can be further evolved into an exact description of the subject in a photo. For example
By recording the perpendicular of side or end, then moving 3-4 paces left or right and taking another view, a 'flip' stereo pair is recorded. 'Flip' stereo is my term for two photos within 3-5 paces of each other of the same subject. Quick eye scans of both images can reveal detail not obtainable from either photo, e.g. detail depth.
The photographic system was further developed into three more sections:
The side and end identification system uses arbitrary features of different rolling stock and was generally determined thus
This is obviously the minimum and was admirably suited for 'N scale' where I tended to stop the detail photos when I had no intention of modeling small detail. Of course for modelers of larger scales, more specific photographs would be required. The establishment of a set photo pattern insures that given minimum time limits and perhaps the excitement of a new find, the basic details would not be missed.
As awareness and knowledge increased the number of additional 'value' detail photographs were recorded for overall interest:
From all these photographs it became easier to view rolling stock as having three 'layers'
With this in mind it was possible to photograph vehicles in service, yet model them for another era by knowing the component and lettering changes. Obviously original prototype photos were superior but in the absence of these or diagrams, at least a start can be made.
Random Selection
To build up an information base for vehicles required more than detail photographs. At 1973 on VR, there were about 300 groups ( called 'classes') for about 23000 vehicles covering passenger, suburban, guards vans, freight and maintenance stock.
With increased awareness, more structural types were found and these were recorded as soon as possible. Within a short time it became apparent that random sampling by percentage might find more differences. A target 10% by class was set, plus any vehicles deemed immediately unusual.
Within nine months the target had been reappraised to capture every Victorian Railways vehicle in service. This was introduced to maximise the effort. In most cases I had to walk past vehicles to photograph others. Why not record the lot then? At this time some 6,000 four wheel vehicles were destined to the scrap heap within five years. A golden opportunity to record vehicles which had an average age of 60 years. This photography achieved the following:
To ensure that as many vehicles as possible were photographed it was decided to embark on field trips. These were intense forays to metropolitan and country rail yards on a regular basis. By persistently returning to the same locations on regular intervals, the cross section of available rolling stock to photograph was increased. Not only were some locations more viable than others but trip routing was selected on the basis of best lighting conditions for each location. Generally trips in the Northern areas were clockwise and trips to the west and south west were anti clockwise. This took advantage of sunlight and yard orientation during the day. Examples of locations and frequency are:
Trips were based on curiosity and spaced sufficiently far apart to ensure a fresh input of stock to photograph. Unusual activity and happenings (tests, loads) took precedence.
The photographic guidelines were:
Field book
When the photography escalated so did the logistical problems:
To assist this endeavour a 'Field book' was established. To quickly set up numbers a special rubber stamp was made. By stamping number blocks onto A5 pages, all number groups could be easily accommodated without worrying about gaps. The number stamp had the following layout:
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
1
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
To represent 199 numbers in a class, the first block shows 01 - 99, the next block is prefixed by '1' to represent 100 - 199, etc. With this system, vehicles could be found within about 10 seconds.
The 'Field book' required restamping every two years due to wear and tear and rolling stock changes. With the start of computer operations in 1984 to assist the hobby, a 16K handheld computer was programmed to store 12,600 vehicles numbers. This was the fleet size at that time. Any vehicle could be accessed in 8 seconds.
Negative Indexing
My start in 35 mm photography began in 1965. Up until 1975, no indexing method was organised. This is fairly typical for those with small numbers of photos. Most items can be found quickly by a perusal of all the photos on file. My method to this time was to store the developed films as negatives and only print photos as required.
Negatives were stored as film strips in chronological order using a system created in 1965. Slides went through three numbering systems until 1980 when a fourth was implemented. The numbering methods are:
Note: Any numbering system requires three functions
With the sudden ramping of the rolling stock photography two index systems were implemented:
Pages were stamped with number blocks. Film numbers were written under the vehicle number to record the location of the negative. A quick view of the film would find the exact negative. This combined speed, convenience and the fact that few photos would actually be 'recalled' for use. As there was insufficient room for more than one photo, a supplementary system was devised. An 'X' placed within the vehicle/film number indicated direction to another "Duplicate Photo Index". This index comprised of 1" x 3" cards specially chopped. By 1985, some 17,000 vehicles had been photographed with some 5,000 duplicate cards. Access to any of these photos took about 60 seconds.
Note: To date, all non railway photographs have not been indexed, say 25% of the collection
To track the detail photographs and to establish a knowledge base, a Detail Photo Index was created. Quite simply this was a card file system that had a card with a vehicle photo glued to it. One side was the photo and the other contained the stamped columns that described the sides and ends. Negative numbers were added to the columns as films were indexed. There was room on the card for other interesting details that were recorded.
As every film was developed and indexed, each vehicle was compared against the current photo cards on file. Any vehicle photographed that varied significantly by lettering or structural difference caused a new card to be created.
This system ran for a brief five years from 1975 and some 1500 detail cards were created. The work was so labour intensive that by 1980 it was stopped. By this time attention was starting to focus on data research.
Data Collection
Very rarely does data collection just happen. It must be pursued. The type of commitment will ultimately provide the key to the amount of data. Time is the other factor. The more time available, the more chances of getting the right stuff.
To assist in the endeavours of data acquisition, the following points are worth remembering:
Data Storage
If research was simple all the information would come bundled in the correct order. Such is not the case. The variables are
So how is all this taken on board and stored? Great difficulty. The information storage path evolved from 1978:
As a reaction against taking notes, the camera became the recording medium. To date the significance of the photographs has not been expanded.
A greatly increased data inflow saw a pressing need for some organisation. Particularly standardisation of observations and jot notes. A cross referenced notes ledger was established. All information or data received or observed was typed up into a loose-leaf folder. Text was black, with rolling stock codes typed in red. Each item was numbered and dated. A separate card file tracked the classes by note number. Published information was transcribed into these notes. This was an attempt to standardise all information into a single format. At this time a single four drawer filing cabinet was being used to store diagrams and other hard information. A second file cabinet was added in 1990.
By 1981, the research into rolling stock data was ramping up as the photography was rolled back. The data flow was involving more specific vehicle histories. With some 17,000 vehicles recorded, the focus on data collection turned to information about each vehicle photographed. It became clear that it was faster to collect ALL information than it was to "hunt and peck" for specific vehicles.
To this end a loose-leaf ledger system was stamped up involving some seven folders of about 200-300 pages each. Vehicles were placed into basic folder groups ( passenger, van, freight ) and classes were alphabetical within these folders. The 'I' wagon class of 16,000 vehicles was a folder by itself. To accommodate the data, each vehicle was limited to 2 lines on a page. Whilst this is very restrictive and cannot easily handle out of order data within the lines, it was the only way possible at the time. In fact, this system merely duplicated the large bound ledgers that retained the original railway data, in use since 1886. Excess data was handled by a numbered appendix system.
The work was labour intensive and required decisions on best data. Much of the histories involve cross referencing which introduces the likelihood of errors. Others in the field had all produced their own versions of this ledger system. Despite the books or the size of loose-leaf paper, it was doomed to failure due to:
With all this work, it became clearly apparent why information is not shared. After actually compiling and writing up all the data, the next step in the distribution chain is a trip to a photocopier. The expense and time invested already plus the extra effort of a copy made sure that information sharing was reduced to only those conducting and sharing within the "group". Anybody else requesting "free" information was discouraged. In most cases the depth of work to reach this point was incomprehensible.
From 1986 I began to access a PC computer. Through 1986/1987 I experimented with DOS commands, DOS files and a program called dBASE III to learn about files and database structures. What was not understood was the complexity of rolling stock information and the implied use of meanings and terms. This was the transition between "static" data and a transfer to "dynamic" information. Typing a letter, text file, or simple data file without manipulation can be considered "static".
It took six months to understand the nature of the data, how it could be used and how a data file could be structured. At this time a 5,000 entry trial was conducted to validate the thinking and programming instructions. The structure was solid but the programming revealed an internal dBASE III "bug". One month later after developing a workaround, data entry proceeded non stop for nine years. Initially, it was only possible to add data. Output and statistical routines were all developed during data entry.
The most surprising facet of this program was how the standardisation of data and the enforced meanings suddenly enabled a host of statistical outputs that had never even been considered initially. The longest routine to develop was the cross referenced output. This took three years and about four attempts before working to 99.99% satisfaction. To date it only fails on three interconnected vehicles that have a complicated history.
An interesting thought with the introduction of the computer program was the fact that despite thousands of hours of data entry, none of the data was visible. With any ledger system, the eyes and brain continue to digest discrepancies in information. This reinforces the urge to 'fill the gaps' and find and complete all the unknown information. The computer just sits there. I developed a routine that was assist in the information question and provide cues for information gathering. With 'rules' and an audit system, the program can quickly show where the information gaps are.
The transfer from ledgers to data file also destroyed another segregated system. Ledgers are logically broken down into Locomotive, Passenger Cars, Freight, Guards Vans and Maintenance Stock, with specific grouping within those categories. Unless a separate data file is used for each group, then another method must be devised to tag vehicles with an identification system to guarantee uniqueness. With the Victorian Railways, the class letter M was used for :
To this end each vehicle was tagged with a two letter ID that ensured when cross reference data was added, the correct vehicle would be tracked. The two extra letters indicated
The first letter here was V for Victorian Railways, VicRail and Vline. For the above examples from top to bottom, the ID applied was VA, VB, VC, VD, VE, VF, VG, VH. i.e., Harris M501 became M501.VF whilst a later Comeng car with the number M501 would be M501.VH. The alpha letter was applied on a first use basis and forms no part of the official history chain.
This methodology allowed every class designation to be used across multiple owners, maximum 36, and within every owner the same class letter could be applied 36 times, more than sufficient expansion capability.
A start was made to index and cross reference all the photos and notes using the "Register" program. As this program was specifically designed for rolling stock data, there was some compromise and hard work to index other stuff like family photos, waterfalls, views, etc. A project was considered that converted text output to HTML links to access scanned photos. The methodology was figured out but no action was taken. The work can still proceed. The method was to access an image via an intermediate ID. This ID became the FileID of FTS.
Utilising all programming experience to date, I began to formulate the File Tracking Program. Whilst existing commercial archival and organisation programs could have done the job, my concern was the time spent compromising on data entry to make them work. At a later date, the program may not function as required. Though slower to develop and "clunkier", FTS has been specifically designed for the job:
The FTS system is an archive indexing system that provides sufficient information to find an image, document, or computer file. With careful standard data entry, most material can be found quickly, years after it was added. Longest search to date is about 5 minutes for something obscure. Any arguments against 'home grown' programs can also be applied to commercial applications, i.e. Garbage In, Garbage Out.
Indexing is rather fast and has been designed for single person operation, unlike archive and museum type programs that employ staff to routinely index material. With focus, a couple of hundred entries can be indexed per day ( 500 images ex CD indexed in 3 days ). At peak file collecting, FTS was logging 50-80 files per day.
With FTS development finished it was time to revisit Register for Windows conversion and the image access within FTS. With a large number of images, I did not think that the time taken to do two lots of data entry for both systems was worth it. There was a concept in 1992 whereby Register could create HTML links to existing images and have this information as part of the history outputs. By 2003, this had expanded to a point where it was feasible for Register to interrogate FTS and return back image and link data. Thus all images and reference material could stored via FTS and all history data could be contained within Register. This would avoid problems of dual entries and the associated change problems. To reflect the Windows conversion, Register development was renamed HTS - History Tracking System.
As Register was a standalone DOS based program, many features including the file manager, text editor and printer controls are no longer required and will be removed.
Distribution
Distribution of the data forms an important part of the research. The biggest hang-up to published distribution is the perceived lack of information. Some are reluctant to finalise and publish until all data is at hand. Ultimately, the reason for collecting is to dispense it. The various methods are:
Guidelines to behavior
Ever wondered why companies and organisations don't like photographers and info junkies? Here's a few
So the following "rules" apply if continued access is required
Digital Photography
In the ensuing years digital photography has gradually made inroads into the 35 mm scene. What has been outlined above still applies but must be adapted to suit the digital environment. 'Film' means digital storage and battery life.
Whilst there is some control over the size of the image ( i.e. pixel width and height ) the best option is to go to maximum resolution unless the close up details only warrant a lower setting.
Perhaps the biggest down side to digital is the inability to enlarge portions of the image sufficient to determine some detail. With 35 mm, it would be a fair comment to say that for most uses 35 mm photographs were recorded on a medium which recorded in a greater resolution than required. In some cases, years after the photograph was taken a small interesting portion of the image (not of interest at the time of the photo) could be enlarged to obtain some information.
With digital resolutions up to 2560 pixels wide/5 Mb chips this is not possible at the moment. So we effectively lose that ability. That is why I stress taking photos at the greatest resolution possible. If the main subject is known, fully framed and is in close proximity then a lower resolution is appropriate.
Conclusion
That's about it. I hope the outlines provided here can assist those make a start with their efforts. I have tried to include all the influences I have discovered as well as all the mistakes I've made over the years.
Good luck