Interview 1 – Future Nostalgia Project
Place: Digital Preservation Lab, Cambridge University Library
Interviewer: Leontien Talboom
Date: 2 June 2025

Interviewer: To get started, you wanted to give a little bit of context.

Interviewee: I worked for a number of years with a company that did agricultural development planning. In 1982 I was asked to spend about a year and a half in an African country as monitoring and evaluation manager when the project was at an early stage in its implementation. Essentially my job was to organise collection of data by a staff of people from the communities of the region.

My job was to design the research, train the staff to collect the data and then to manage the data. The colleague who had written the proposal and estimated the budget for this project, which was supposed to last 15 years or more, had put in an allowance of £5,000 for timeshare processing in UK of data collected on site. When I was asked to join this project, I was aware that there were microcomputers that would provide an efficient way of organising the data on site. Processing the data at the project headquarters would also mean that the data would be readily accessible by the people who would come after me. My predecessor put all the numbers for the various things that had been recorded into financial ledgers, which I looked at, but I was never going to use them. And his analysis was restricted to adding up all the numbers in a column and saying the average of these is X. He approached the work with a very different mindset to mine. To me, understanding the agricultural practices of the region involved information about the diversity of what goes on there, rather than to summarise the farming practices as an average number in various aspects.

I suggested that the available budget should be used for a microcomputer. The company was very brave in going along with my proposal. I had worked for them for about three years by that point and they had seen that I'd collected data and organised it in other contexts. I selected a CP/M-based microcomputer with a couple of floppy drives and a dot-matrix printer. There were two crucial components. One was called a power bank, which was a large box containing a number of 12 volt sealed lead acid batteries. The other item was a portable generator. Now the problem with a generator is that the voltage coming out of it may be quite variable. The approximately 230 volts coming out of the generator went into the power bank where it was transformed down to 12 volts. At the other end of the power bank it was turned back into 230 volts AC with what's called a modified sine wave. This meant there was relatively consistent energy going into the back of the computer, which minimised the risk of damage to the computer because of unsafe voltages.  The whole thing worked fine for my year and a half there. After I left there was a fault on the motherboard. But, by that point, people on a related project had already ordered a couple of the same kind of computer. The people that I'd worked alongside had seen that I was able to write reports while overseas and put graphs and tables into those reports. These were tasks which had previously involved the UK office.

After I retired, I had a chance to work on a freelance basis on a project in the same country. That was in 2014-15. People I worked with had connections with people working at that time on the project I had been involved with in 1982-83. I sent out an offer. If I could have on loan all the floppy disks collected by the project, I would read them and give the data back to the project staff on a modern storage format. I was sent 800 floppy disks. My motive for making the offer was mainly based on the fact that the evaluation of that project never happened for reasons to do with international politics around the 1990s. My aspiration for that effort was that a post-graduate researcher, for example, with links to that area, could access the data and past reports that I'd recovered from those disks and use the information as the basis of a PhD to carry out the evaluation that had been planned but which hadn't happened.

At around the same time, in 2015 I think, I visited the Centre for Computing History at Cambridge. I recognised in the collection a computer like the one that I had persuaded the  company to buy on my recommendation in 1982. I made a similar offer about reading floppy disks in the museum's collection. The founder and director of the museum took up my offer. 

Interviewer: I've never met the founder of the museum.

Interviewee: Basically, the director said: if you look around and find some disks, you're welcome to take them away, read them, bring them back and at some stage we'd like to have the data you've read from them. So that was the start of my involvement with the Cambridge museum as a volunteer.

Interviewer: How many [disks] have you found? 

Interviewee: Now that's a tricky question. I'm going to mention some of the software tools that I've used. One of these wonderful tools is called TreeSize and I've licenced this for myself for more than 20 years. It's a product of a company in Germany called JAM-software. One of the great things about that software is that you can select the file type, for example RAW or IMG or DMK or IMD. The software it will show you those directories that contain files of that type, and how many files there are. I used this to export, for example, what it told me about the number of DMK files. For various reasons I'll come onto later, that file type was one of my favourite ways of doing a first image of a disk. Very often, I then  converted the DMK files into IMD [ImageDisk format]. An advantage of IMD disk images is that the user doesn't need special hardware to be able to read them, whereas the DMK needs to be used with a Catweasel device which you may or may not have heard about. 

I built up spreadsheets where I had for example disks used with, let me think of a suitable example, a 'Cromemco' computer, which was popular an American manufacturer in the 1980s. I was able to put into the spreadsheet the number of DMK images had in that category and how many IMDs. I decided to take the bigger of those two numbers. And that could be repeated across all of the manufacturers of the computer systems. In each case, I took the larger of those two numbers and that came to about 8,000. Now, I have never been back to validate that and I can only say that it is approximately in that area because it had become a large and complicated collection. I am conscious that there is a distinct likelihood that I've made duplicate copies of the same disk image because my arrangements for filing this stuff sort of grew and developed as the project went on. And so I didn't have an architecture at the beginning that everything would just slot into. At the time I assembled the collection, I was familiar with the various collections of disks. I reached a point where I thought 'I've devoted enough effort to this'. In other words, if somebody then looks at this archive of data, they'll probably have to make a bit of an effort. It is appropriate that they make the effort rather than me, because that way they will learn about what is in there. And if they come across duplicate files, I'm not embarrassed about that.

Interviewer: No you shouldn't be. Yeah it's like it's ineffable isn't it.

Interviewee: Well one of the other features of TreeSize is that is able to find duplicate files for you. The user can select which of several various algorithms is applied to produce a signature that's written out as a hex string (a 'hash function' such as MD5 or SHA-1). You can sort on that hex string and you can identify where there are directories containing multiple copies of a file. So in a sense I had the tool to be able to deduplicate. However, I didn't want to destroy the coherence of any particular directory. I didn't want to delete half the files in this directory and then leave another directory with the other half. It was better just to take up a bit more disk space.

Interviewer: Yeah, exactly. To make sure it's all there. 

Interviewee: Perhaps I can elaborate a bit further. In late 2023 I gave a copy of that version of the data to the chief executive who had taken over at the museum. I had written two copies of the complete data set on a laptop hard drive. Each copy on that drive amounts to 160 gigabytes because it's not just disk images. I also did research into the background to the floppy disks. For example, I mentioned the name 'Cromemco'. Well, there are lots of American repositories where you can find other disk images associated with the computers that were popular in the US, as well as manuals and other documentation. I gathered together data that were relevant to the various sets of disk images that I had read. So in a sense, the disk images had become a relatively small volume of data as a share of the whole exercise. 

I had started out thinking that my key role would be to find boot disks. The museum's policy is still to have computers switched on and a chair in front of the computer, so visitors can sit and manipulate the computer. What I reckoned was that visitors would put disks in the wrong way around or they'd pull out a disk that had been used to boot the thing and there would be a demand for spare copies and the means to make additional copies. But actually, that's only happened once or twice in my experience. 

What I quickly realised was that many disks that had been donated, together with the hardware that had written those disks, contained user data. For example, there were company financial records, children's homework and all manner of files. The insight that I obtained was that this user data, across the whole body of the different computers, and the different years in which they'd been used, and so forth, provided potential raw material for somebody who wanted to do research into the sociology of the personal computer era. Who were the people using personal computers in the late 1970s? How did that change during the 1980s? And how did the profile of users change further in the 1990s? I would say that my interest drops off rather at the end of the 1980s. That's a period when things were very diverse before Windows and the like.

Interviewer: Like once Windows came, it was not as cool anymore. Yes exactly. It goes to single flavour, doesn’t it?

Interviewee: One of my initial insights into user data occurred when I bought for myself a Sirius One computer that was being sold on eBay and which came with a hard drive. When I read the hard drive, I realised that the original owner of this machine put all of his trust in a single software package, a spreadsheet package. He had used this spreadsheet for his company accounts, but he also used it to send messages to his family about a holiday. One can understand that because software packages in the early 80s cost something like 300 pounds. Why would he spend another 300 pounds to have a word processing software package if he could do what he wanted to do using a spreadsheet? 

Interviewer: That's so interesting. 

Interviewee: It is that kind of insight into that person's approach to the computer that was now in my possession that made me think that the raw material could say many things. At a very basic level, when I was working on the computer that I had used in Africa in 1982 and 1983, the files that I wrote onto floppy disks were miniscule, a couple of kilobytes. They were tiny because one then made backup copies of those tiny files.  I was faced with putting in the data for five different locations where my staff had gone out to record observations during the agricultural season. It was best to do each of those five separately rather than put all of them together because you never wanted to put too much data together in case you lost it. Given the widespread adoption of hard drives and so forth as the 1980s went on people were much more confident that they were not going to lose the data and the files got bigger. As I say that's a very basic observation but it tells you something relevant about how much time people devoted to the data that they worked with. And with a hard drive, the user's productivity could go up dramatically. 

Interviewer: It's really interesting to hear as well that, and I think that's a piece of research that is really missing within my community as well, is what did people actually use personal computers for. Because we get a whole bunch of disks or like stuff into the library and I should probably know, sometimes there's notes on them, sometimes there's not. And just having a better understanding of the actual use of personal computers in that time period would probably help us understand what we're actually looking for. Like your example using a software package that does spreadsheets, why would you buy another software package? That is really expensive when you can use the spreadsheet package. Whereas like if we don't really have that context, we're just, it's unclear why there would be personal data in something that would be more seen as a business package if that makes sense. There's something really interesting there.

Interviewee: A further insight, one of the things that I was very happy to do for the museum, on occasions when somebody quite a long distance away had offered to donate an interesting personal computer, I was ready to go and meet that person and I always thoroughly enjoyed meeting these donors and then bringing the computer back to the museum. One case that really stands out for me is somebody who had used his own money when he was in a small team at work and the manager of this team had gone round the corner to a computer shop and had bought several machines of the same type and thereby qualified to pay a little bit less for each of the them. They were owned personally by each of the team members, and so this person had kept it at the end of the project. The boot disk that had the name of the company that had made the computer could hold 800 kilobytes of data whereas the operating system needed around 100 kilobytes. The owner had filled the rest of the capacity of that disk with reports, data files, all manner of things. He was reluctant to go and buy another box of floppy disks when the disks he had already could accommodate the files.

Interviewer: This explains why we have a boot disk with personal data in our collection.

Interviewee: Well exactly. 

Interviewer: But it makes total sense doesn't it? Yeah.

Interviewee: For that reason the distinction that I thought might exist when I began, here is a boot disk, you know, that can be given out to some member of the public who's acquired one of these computers but doesn't have a boot disk. Well, no you can't do that without having checked all of the files on it, which was a step that I wasn't prepared to take. 

Interviewer: That is so fascinating. Because it's that, isn't it? Like, it's just humans being humans. It's the same with the invention of stuff like the flippy disk. That wouldn't have been around if people weren't like, well, 'there's another side to this disk'. Why not just use it? And there's probably many, many of those examples. But yeah, boot disks is another one, isn't it? They've got more room. And if there's nothing else on them, why not just use it?

Interviewee: Moving on to another part of my personal story, if I may, is that I made a distinction between what I call special collections of disks and a general collection. To be in the special collection, things had to be a body of data that had been written over a period of time using different computer systems, different sizes of disks and so forth, but they all related to that entity that had created them. So one of those is Intel. I became very interested in Intel's 'development systems' of the 1970s and 1980s. These computers were commonly sold or leased to businesses that were coming up with ways of applying computer methods to industrial processes. To describe an example which I was told about by somebody from whom I bought one of these Intel development systems, his company was commissioned to come up with a way of filling bags of potato crisps. To outline what he told me about this, the cooked potato crisps came along a conveyer belt and they then dropped into a number of very sensitive scales and the computer worked out which of those 12 sets of crisps, when dropped into a chute and into a bag, would give exactly the right number of grammes. Now the machines that were developed by this company were capable of filling 150 bags per minute. That's two and a half bags every second if I've remembered the arithmetic I did at the time. And so this was a major step forward in that industrial process in terms of the consistency that you could trust that manufacturer's bags of crisps to have the weight that was stated on the package, but also to be able to get the volume of throughput. Now, Intel's development systems were about the size of the table in front of you [points to side table] and part of the work needed to be carried out at the industrial site. 

Interviewer: Oh, interesting. 

Interviewee: The computers used eight-inch floppy disks. And so, there were various barriers to this process working for contracting companies. Large companies would provide their in-house staff with these same Intel development systems to do software development in a similar way. 

Interviewer: It's that period before it that is more interesting isn't it? Where everyone was kind of like trying out different things and doing it slightly differently.

Interviewee: But actually making enormous strides forward in what they were able to do.

Interviewer: But also just like I don't really know how to explain this but also people just experimenting and trying stuff. Like people being quite, I would say, like very happy to adapt. So not only with like the different floppy disks, so again, like flippy disks and inventing or not inventing, but like using that type of stuff, but also just happy to adapt their own computers or try different things. Whereas like I find with Windows and Mac, especially Mac coming around, it's not really encouraged in the same way. There's not that kind of interesting relationship that people have with computers.

Interviewee: I have a friend who is now well into his 80s. He was producing a book on his personal passion project. I won't identify it further but he's a very talented writer and photographer and he did literally the entire production of a book and he then found a printing house that was able to print it with high quality colour. He is by nature a one-man band and having a Mac with the page layout software that was available to him in the early 1990s, he was able to do things his own way and produced absolutely wonderful books. 

The whole book was laid out immaculately with a very logical structure related to the nature of that person, a very tidy, orderly kind of person. He was able to do the whole job that 10 years earlier or even 5 years earlier you'd have needed to send to a specialist where they would have custom software running on minicomputers with capital costs which were huge.

Interviewer: It's interesting, isn't it? Also going back to the collection that you made for the museum and you saying it's 160 gigabytes because it's not only disk images, it's also that information that goes around the software. So just out of interest, I find that really fascinating because it's like around kind of that context building because you kind of showcase that, of course it's good to have a disk image of a specific system, but if you don't really have a manual or any other information with it. The data is not unreadable, but it's nearly unusable. It isn't able to tell you as much. Have you got any tips in like finding that type of material?

Interviewee: Well of course Google. 

Interviewer: Google's your friend. 

Interviewee: To some extent the existence of bitsavers.org and archive.org. These are massive collections.

Interviewer: And finding that information in those types of places because I did see that the museum does make a number of manuals available. 

Interviewee: Yes, that's right. They actually have a lot of paper in the collection.

Interviewer: There's a lot isn't there?

Okay, so just a little bit of context for you as well. We don't really have a problem with getting raw disk images. I wouldn't want to say easy, but it's a lot easier for the more modern stuff. So like when Windows comes around, when Mac is around. Mac has its own problems with resource forks and stuff, but still, like it's not like the pre-1990 stuff, which is just a whole range of stuff out there. And that's the stuff that we struggle with as a community, but I think we're also very worried about because that's the stuff that is slowly coming into our collections. So a lot of people who were active in that time are either retiring or they're starting to pass away. And that means that we get these collections into our libraries or institutions. 

And yeah, I think we have to be better at getting the context of these types of systems together. Because otherwise, even if we get the disks in and we can make a flux stream, we can't really do much with them if we don't have any more around that. And it's really fascinating that you mentioned collecting manuals and other materials around the disk image. Because yeah, I completely agree. I think we're losing that information.

Interviewee: I mentioned to you that currently one of the things I am doing for which I go to the museum is taking the hard drives out of Mac computers to read the contents. 

Interviewer: Yes, you were saying that.

Interviewee: I have some tools [ZuluSCSI, BlueSCSI, etc.] that can extract the entire file system from a SCSI hard drive without needing it to be in a computer that is operational. 

Interviewer: That's cool. 

Interviewee:  What I'm doing basically is having a very brief scan through what's on there and then trying to identify 'was this used as a family computer or a personal one?', or 'this one shows signs of having been used in a business or public services'.

Interviewer: How do you make that distinction?

Interviewee: Well, for example, supposing there are directories with the names of several people. That is one thing that could suggest that it's perhaps a family computer. But I always look at a hex file of the entire file system. And there was one particular case where I eventually persuaded the CEO that that hard drive should be deleted, that all of the data should be wiped. As I scanned through I kept seeing reference to 'benefits'. I realised that this computer, when it was first bought, it had been used in a housing association, but had later on become a family computer. All these references to benefits were letters that had been written to tenants who needed to be claiming all the welfare benefits to which they were entitled to be able to pay their rent. The person who had donated it to the museum perhaps believed that all the files relating to tenants had been deleted. But when files are deleted, it's just the directory entry that is marked, and the data may remain in the sectors where it had been recorded.

And so this hard drive had on it names of people, all of them long gone because this is probably from the 1990s. But nobody nowadays, particularly nowadays, is truly long gone because we all leave such a footprint of our time. I argued that the museum ran the risk of being embarrassed in the event of somebody coming across this material. However, it might happen and then the museum might be regarded as not having applied diligence about the nature of the data that it had received with donations. And that comes up both with family systems and also business ones. It would be a tragedy to go around wiping every disk in the museum's collection in order to guard against that potential risk. And so what I'm doing is building up a repository of disk images which then can be held securely. I envisage a policy of reducing each hard drive to the operating system or something like that. So that the computer could be restored to working order, but it would not have any personal information.

Interviewer: But not have any sensitive or personal data on it, interesting. So that you've got like a clean install of whatever the operating system was.

Interviewee: So I haven't done the second phase of this and really my preference would be that other volunteers with particular interest in that generation of Mac computers would do the re-installation. I recognise that my small steps towards learning about the whole Mac ecosystem have not got me to the point where I really know enough to be able to do this with any degree of skill. Similarly, I decided a long time ago that I was not going to try and get into the Acorn ecosystem because again, there are countless other people who know all of the peculiarities of that. My bias has been towards computers that were used for professional purposes. I'm not very interested in people's homework or their attempts to do graphics, you know, to do paintings and so forth on computers. 

Interviewer: Yeah, all the different operating systems and formats that you can find for floppy disks is something that has really, I didn't realise that it was such a, I wouldn't want to say problem, but that it was so diverse. It's a mess. It is a mess, isn't it? So we've got a floppy controller in that machine, which is an FC5025.

Interviewee: I tried to buy one of those and I was told he wouldn't send one out of the United States. But it's scope nowadays is really rather limited, I believe.

Interviewer: It's very limited and it's very limited if it's like it's great. If you've got anything that's MS-DOS based, but as you probably know, here in the UK, the market share for those types of computers wasn't very high at the time. So for us, probably if it's an institution similar to us that is based in America, they could get far more stuff copied or like imaged with an FC5025. But for us, it's not. It's very limited. Which is again, something that not a lot of people talk about how there's quite a geographical disperse of operating systems. 

Interviewee: The original floppy-based IBM personal computer wasn't marketed in Britain and Europe until well into the early 80s, which is how computers like the Sirius, which I talked about having had one at one time, which had a number of technical features that made it better than the original IBM personal computer. The Sirius One was marketed by Apricot Computers, which was a British company which produced various machines which fitted well into the way that small businesses operated and were sold at prices that families could afford. 

Interviewer: So fascinating, isn't it? That a lot of that comes down to marketing and like, it makes sense, doesn't it? Cause it happens nowadays as well. Like we basically buy what's available on the market and you only get a small number of people who are that passionate and then buy outside of that. But yeah, it was the same in that time, wasn't it? I didn't know that about the IBM PC Floppy Disk system (not) being marketed here until much later. That's really interesting. Do you know why they did that? 

Interviewee: Well presumably demand in the United States took off much more rapidly. The IBM personal computer wasn't their first attempt at producing something that was much more compact than IBM had a reputation for, but it took off. 

Interviewer: Took off quite a bit. It's so interesting because you can really see a difference between, like if I talk to my American colleagues, just the difference in disks that they have. We have such a wide variety of stuff. Also, depending on institutions where you sit, so I've got a colleague who works over at Churchill Archives Centre. That's a lot of political data and a lot of government data. So some of their disks are for ICL machines because they were used in government.

Interviewee: I am happy to help with that. I bought an ICL machine which the museum was selling as a duplicate they didn't need to keep. 

Interviewer: I will say to Churchill. And it’s fascinating, because we mainly get stuff from personal collections and literary archives, so like writers or poets, type of stuff. That means that our data and the systems that people used very different from other institutions. 

Interviewee: Like the Amstrad?

Interviewer: Yes, we've got disks for the Amstrad. They are their own kind of pain, but also a lot of fun. Thank you for providing a bit of context before starting with the sections [refers to interview questions].

Interviewee: Of course. 

Interviewer: So, the first section of our questions is around sourcing and maintaining equipment. I know that you mentioned buying an ICL machine at the museum. Yeah, the sales that the museum do are great as well because you've got an ICL machine there.

Interviewee: Which I believe was a better example than the one they kept. 

Interviewer: I've helped out with a number of those sales and yes, there's some real bargains to be had. But yeah, like my main questions would be, and also again to say any of this type of information will probably feed into a resource that I'm creating with the community around like how we should be capturing and like sourcing materials. But yeah, anything around, if you for example find specific floppy disks, and I know you're associated with the museum, how do you then know what hardware you need to actually read it? 

Interviewee: Well, the first clue is the manufacturer's label which most floppy disks have in the top left which will tell you how many tracks per inch and whether it's single sided or double sided. The box the disk came in might have additional data on the format but you can generally select a drive that's going to rotate at 300 rpm or 360 rpm just based on the type of disk, although that is not an infallible guide. So to answer your first point there, it is my own equipment with a few exceptions. One of the exceptions is that a visitor from the United States actually sent to the museum one of the early AppleSauce devices. 

That unit was temporarily passed to me, but I have not used it. That comes back partly to the fact that I felt that there would be other volunteers who were more knowledgeable about Apple systems. After some time, I returned it to the museum but they wanted me to take it away again because that way they knew where it was. 

The Kryoflux is a device which I would absolutely not recommend to you because the team responsible for it appear to believe that a museum and any public body has got millions of cash backing it. They thought they would make a lot of money out of licensing it. Now I have one and I used it in my personal capacity. So in other words there was consistency between what I was doing with it and their licensing terms. 

Interviewer: Yes, that licensing is a bit of a discussion in our community. There's been people who've been quoted several thousand dollars to get a licence and like you say, expecting we have millions. We really don't. 

Interviewee: A point I wanted to mention here is that the Kryoflux team produced videos. One of the videos was about how to modify a particular Panasonic floppy drive to be able to access tracks minus one and minus two. This involved physically cutting away part of the body of the floppy drive. It's something that had I ever wanted to do it, I would have been capable of doing it. And in the case of one set of disks that I read, when I sent a few image files to a friend for him to work out how to access the files, he said he couldn't find the directory on any of them. Essentially, the floppy drive on which they'd been written had not been aligned correctly. When the floppy disks were written, they used a part of the floppy disk that was coated in oxide, so it was capable of recording the data. But if you read it on a floppy drive calibrated in the standard way, the directory tracks were beyond track zero. 

Interviewer: Oh wow, okay so that's actually really interesting. I didn't realise that. It makes sense.

Interviewee: The computer which had been used to write those disks is in the museum's collection and so the direct way of getting around that problem would have been for me to open it up, take out the drive, put cables on it and so forth and then read the discs with the drive which had track zero not in its proper place. 

Interviewer: It’s seems that the main thing is to find the directory. Like are the directories the main thing that you want because that's where all the information is?

Interviewee: Critically important, yes.

Interviewer: Would that be your first thing that you would recommend? So someone comes to you and they would say 'I can't find the directory', would your first instinct be try this?

Interviewee: No, what I would do is read it with a Greaseweazle. On one occasion I asked the author of the Greaseweazle firmware and software to explain how can I use this device to produce the same results that I had been able to get with a KryoFlux. He wrote back and explained how to write a command line to produce the same type of files. The key issue in this instance was that I wanted to create one file per track and side. And each of the files written in that way is the result of multiple passes over the track and side of the floppy.

And so these days the Greaseweazle now does everything for me that the KryoFlux was able to do. And what I also do from time to time is use the software that was developed for a flux waveform reader called FluxEngine. The software available for the FluxEngine knows about some non-mainstream sorts of floppy formats, such as 'hard sector' disks. It can be used with the Greaseweazle hardware as well as the FluxEngine device. I think it also knows about Apple disks which present problems normally.  Being able to use both types of software provides much wider scope than if I was only using one of those. 

Can I Bring out one of the exhibits that I brought? 

Interviewer: Yes, please do. 

Interviewee: This little device [FloppyControl Arduino Due v1 2019 by Josha Beukema] is designed to be installed on a type of Arduino with an ARM processor. The floppy cable is connected here [points to 34-pin interface]. This is the clever bit [points to small red PCB] because below the heat sink there is a micro stepper and it can send commands to a 3.5-inch drive that move the head a fraction of the distance between two tracks. The name of the person who developed this is printed on the PCB and you're welcome to photograph it if you'd like to. 

Interviewer: Yeah, I want to. 

Interviewee: Wires going to the drive circuitry would be commanded by the little chip under that heat sink. I believe his focus was on games and other relatively recent stuff, but only on 3.5-inch disks. The notion behind this was brilliant. And it's great pity that, because of his poor health, he didn't finish off the design. Anyway, his second brilliant idea was to use an oscilloscope, which is what people use to read waveforms, to read a single track where there was a sector that you could not read correctly because some of the flux transitions didn't make tidy waveforms. He wrote software that allowed you to look at the waveform and to identify places where the waveform didn't go high enough to reach the voltage level that registered as a 1 rather than a 0.

Interviewer: Wow! 

Interviewee: And then you could find a similar waveform, cut it out from there and paste it in to replace the problem and then the software would tell you that it's now able to read that sector. In other words that the CRC, the cyclic redundancy check, which is derived from reading the flux transitions encoding the data, matches the CRC that was recorded on the floppy disk when the sector was written.

Interviewer: That's such a cool project.

Interviewee: Yes, I have not been back to him for a long time. I know of people in Britain, one of whom I've met and talked to about reading disks at the Cambridge Museum, who used exactly that approach to be able to overcome a problem in reading a really critical part of a disc. In principle, the use of an oscilloscope to capture a track should be capable of being adapted to use on 5.25-inch, 8-inch or whatever. 

That's one idea that did not go through to a finished product, but I was fascinated to have learned about the Floppy Control App. 

Interviewer: Yeah, it's so cool, isn't it? And it's so nice to see that people make such interesting tools. I would love to see more of that happening. because it's fascinating when people do that. So I can see also that when you're sourcing material, or like hard to find, stuff. That eBay is your main place to go [reading from notes provided by Interviewee].

Interviewee: I'll give you one example of eBay being a useful source. I met somebody who had been a great fan of the Amiga computer system and the Catweasel device which I've mentioned. He was selling the Catweasel [an early example of a flux transition capture device]. By jumping in quickly with an offer I was able to get it at a price that was perhaps a fraction of what it might have sold for today. The Catweasel I got was not the latest and greatest version. The museum has one of those and, if I'd wanted to borrow it, I'm sure they would have loaned it to me. But it means that throughout most of the period that I've been reading disks which were donated to the museum, my first step typically has been to use the Catweasel, not because it's hardware is outstanding, but it's software is very smart. And is now is a good moment to talk about it.

Interviewer: I would love to know about the software. So just to give you a bit of context, we've got the FC5025. It doesn't really do what we want it to do, so we've stopped using it. So we now use a Greaseweazle. And we use both the Greaseweazle software and the Flux Engine software that you mentioned. I have also installed the Applesauce software onto our Mac laptop but we do not have an Applesauce floppy controller you've probably seen this they're really difficult to get your hands on. So that's basically what we use and I’m aware of the Catweasel and I've heard multiple stories people talking about.

Interviewee: It's like hens teeth [impossible to find these days]. Anyway, so just briefly to summarise. The Catweasel goes into a PCI slot, so it's not able to be used with very modern computers but there are lots and lots of suitable machines. It's not dependent on your machine having an ISA slot. It is capable of being linked with a second cable onto a floppy drive that is built into the computer, which is a feature that I've never wanted to use. Anyway, what is clever about it is that you can put in a 5.25-inch disk and it will work out whether it's single sided or double sided, whether it needs to double step.

Interviewer: It does that all by itself?

Interviewee: By itself. And so you can put in a disk about which you know very little and where, you can write onto a floppy disk what your disk writing software wants to put there. This device works it out.  It does so regardless of the manufacturer's label, because basically a floppy disk is a blank canvas.

Interviewer: That's really neat. 

Interviewee: It also copes with the not very common, but well worth knowing about,  arrangement where track 0, side 0 is written in single density encoding (also known as frequency modulation) when track 0, side 1 and all the rest of the disc is written in MFM (double density) encoding.

Interviewer: Why? 

Interviewee: I think the idea of this is that in the early days any floppy drive would be able to cope with frequency modulation. So it would be able to get started on the disk. But of course FM was dropped in the mid-80s and the floppy controller hardware didn't even know about it (when the functions needed to control a floppy drive were integrated in a single chip). So you were left with only MFM capability. But the Catweasel was essentially a flux reader. Under Linux software it can be used as a flux reader. I haven't ever done that, but it enables the Catweasel to read some really weird and wonderful encoding arrangements like modified-modified frequency modulation ('M2FM') which was used by Intel and Hewlett-Packard for a relatively brief period. You can read M2FM disks at the level of flux transitions using a Greaseweazle but you won't be able to make sense of the data you've read without specialised software. The Catweasel is not something that I'm frequently using nowadays. However, I'm not really using anything frequently nowadays because it's very difficult for me to find the kind of disks I like at the museum. The approach in recent years has been for disks to be put in storage boxes as they come in. Acid-free boxes and all of that kind of stuff So, I think I'm often not even aware of them.

Interviewer: Yeah, so like it's not as easy to retrieve them anymore. 

Interviewee: Whereas for a number of years I was able to look in boxes and was finding hundreds of disks. 

Interviewer: I'm pretty sure they're still there. But that is interesting about the Catweasel. That's really cool that it can actually know what type of disk has been inserted or like at least give you quite a lot of information. 

Interviewee: One of the other useful things about the software is that you can put various commands on the command line, because it's DOS piece of software, MS-DOS. For example, you can tell the Catweasel software how many times it should read any track where it finds a bad sector. As standard it will do up to six reads of a track where the first rotation showed a bad sector. Normally it is able to resolve any error within that six. But the user can increase that to a larger number which can sometimes archive a successful read.

Interviewer: That's cool.

Interviewee: A further feature is that you can give a name to the disk image, which matches the name written on the label and photographed previously. The file type is a 'DMK', that's named after D.M. Keil, who wrote software for the Catweasel to go with the TRS-80 PC. The user can also write on the same command line that same file name with the file type of '.log'. Everything that appears on the screen while the Catweasel is reading the tracks will be put into a text file. At the end of the reading, the screen summarises how many FM sectors has it read, how many sectors had MFM encoding and how many were RX50 sectors (a Digital Equipment Corporation proprietary format). Those three are the types of encoding that the Catweasel software is able to recognise.

Interviewer: So you get a log file alongside which you read. That's really nice that's something that we like too.

Interviewee: Do you know about WTee.exe?

Interviewer: No.

Interviewee: There is a document about the use of the KryoFlux for digital preservation.

Interviewer: Yes, the Archivist's Guide.

Interviewee: That's right. And that refers to this little utility, which I think originally was a Unix/Linux tool, but it was translated for use with DOS. So you do your standard KryoFlux command line and at the end of it you then put a few things that tell it to write a log file. Now the log file you get with a KryoFlux tells you about how many flux transitions it's read or something like that, which is not really terribly useful because the core of the issue is how many sectors it was able to read and to make sense of on every track and side of the disk.

Interviewer: That is interesting. Are there any tips that you would give people if they're looking for stuff, say online or on eBay, where they would have a good idea. Because like, of course, it's a second-hand marketplace. So sometimes people would list stuff because they find it somewhere in, you know, their, I don't know, their parents' garage or someone's garage and they just list it. And you don't really know what state it's in.

Interviewee: Yes, that's often a problem. 

Interviewer: Do you have any tips on picking the right listing? Or is it always just a bit of a gamble? 

Interviewee: I'm one of the virtuous users of eBay who has never had negative feedback. 

Interviewer: Nice!

Interviewee: And so if I'm selling something, the fact that I've got 100% positive feedback sends a signal to a potential buyer. Whereas if somebody's new to the task or they've only got say a 92% positive feedback record [...]. I have a policy that I won't even look at something when the seller's feedback is 95% or below. 

Interviewer: Interesting. Because I've also, so we've bought some stuff from eBay, as you can imagine. And we've always, I've always bought stuff where there's a little bit more of like specification in the description. 

Interviewee: Currently I've got a large collection of floppy drives, mostly ones I have removed from PCs I have owned at one time. One of these days I must get around to selling most of them. But if I do that, I will describe the tests that floppy disk drive has been put through and which it passed to show the buyer that the money I'm asking them to pay for this is going to be money well spent. And I do have a few floppy disk drives that I would be reluctant to sell because really they're only fit to be used as spares. 

Interviewer: Is there like, so on that topic of like floppy disk drives and like that type of stuff, are there specific eras, brands like that you would say, yes, go for these?

Interviewee: Well, for 8-inch drives. I had a couple of half-height NEC drives which were great in that they didn't need the weird voltages that the early 8-inch drives needed. I wasn't happy about them and eventually sold them. When I had bought them, I was looking for an 8-inch drive that had a motor that directly powered the spindle rather than having a band connected to a motor where stretching of the band, or irregular rotation because of this flexible band, meant that the disk was not spinning at quite the right speed to be read reliably. But having said all of that, my favourite brand of 8-inch floppy drives is Shugart and they all have one of these rubberised fabric bands. With 5.25 and even for 3.5 inches TEAC is the brand which I go to every time.

Interviewer: You're not the first person that has mentioned TEAC. Because they seem to be the most reliable. Like our, favourite floppy disk drive is a TEAC drive. Never lets me down. 

Interviewee: I know, 3.5-inch as well as 5.25.

Interviewer: Yeah, they seem to be the most reliable, which is quite nice. It's nice to get that confirmed as well, because I use them and I'm always like, you're my favourite, but I don't know why. And then, yeah, it does make sense, doesn't it? 

Interviewee: One of the other things that I've mentioned, if you're looking at the page listing your questions together with my comments, is that I have a couple of FDADAP, which is a device which you can put between an 8-inch drive and a floppy controller on a modern computer that only has a 34-way cable.

Interviewer: Oh, that's nice.

Interviewee: The FDADAP translates the signals between the various wires in the 50-way cable to be output on the appropriate wires in a 34-way cable. And at the end of that line I've said there are also similar devices. There's a Frenchman who has made a device that does essentially that. And I also have a connector which allow you to have a 5.25-inch drive on the end of a 34-way cable, but in terms of the signals that go back to the controller, it is regarded as if it is an 8-inch drive. In terms of being able to get the drive and media, high density floppy disks which rotate at 360 RPM have exactly the same overall structure as an 8-inch. When it is a matter of reading the disks, clearly you've got to use an 8-inch drive.  But if you want to operate a machine which originally had an 8-inch drive, you can use a 5¼ inch high density drive with the appropriate media and with one of these devices along the way to that drive. So in other words your computer would actually have a 50-way connector on it. 

Interviewer: Wow, that's really cool.

Interviewee: Then the 50-way cable ends with an adapter and then it's a 34-way cable that is connected to the 5¼ inch drive.
 
Interviewer: Nice. I didn't know of the existence of that. That's sounds really useful. 

Interviewee: I'm glad to find that there are one or two things that I know that you haven't known. 

Interviewer: There's many things that you know that I don’t know.

Interviewee: The whole idea is to pass on these various snippets that I have gleaned. 

Interviewer: Because that is the thing, isn't it? Like it's otherwise all that kind of information gets lost and it's like that's something I can always remember going to the museum. We see Amstrad drives and the 3-inch floppy disks. I didn't realise that sometimes they get a little bit stuck like the tape gets so if you click them they fall into alignment again and you can read them but if you don't do that they're unreadable. It's just like one of those things that Adrian showed me right where I was like, "Well, if I didn't know that, I would have just assumed that our disks were just unreadable."

Interviewee: Can I pick up that point? 

Interviewer: Yeah. 

Interviewee: Because I've handled a number of these Amstrad 9512 machines. Because of my approach I take the floppy drive out and I connect it to the (AT or ATX) power supply that is powering the PC motherboard and the other stuff I've got. That is a very bad approach because the Amstrad power supply limits the amount of 12- and 5-volt power going to the drive. I think what I've done is to burn out components. It took a long time for me to know about this and I burned out a few Amstrad 3-inch drives. 

Interviewer: Yeah, exactly.

Interviewee: I've been in contact with somebody who thought that he had identified a particular, absolutely miniscule, component (you could put 20 of them on a small fingernail), a tiny component which is there to protect the rest of the circuitry if it gets too many amps going into it, that thing will blow but it means that the whole thing... 

Interviewer: It's not protected anymore. 

Interviewee: No. The rest of the circuitry on the drive is protected because that particular thing has to pass the current through. If you try to squeeze too much current through that thing then it will blow after which the drive is not able to operate. 

Interviewer: Cause that is, those Amstrad drives in general were a bit of a faff to get to work because they've got a 26 pin interface. Who's decided that that was a good idea?

Interviewee: It is quite easy to make an adapter.

Interviewer: Yeah, we played around with that. But it is one of those things and it's a 12 volt. But if you were to only look at the four pin, looks like a 5-volt.

Interviewee: You have to swap the five and the 12-volt wires on the four-wire connector. 

Interviewer: It's like, there's so many things to keep in mind, it's fascinating. 

Interviewee: Go there at your own risk because you're right, you’re likely to damage something.

Interviewer: We didn't fry anything. It was all okay. But yeah, I was very worried that I would either break the drive or break the Greaseweazle in some way, but we were all okay. 

Interviewee: The thing about my collection of floppy drives, one day I managed to focus my attention on them and I tested them in turn and used a particular trusted disk. I made a sheet of labels where I had the serial numbers which I'd taken from photographs of all of my floppy drives. I then built up a spreadsheet which recorded that this drive passed this test but it failed that test. So in terms of an inventory, that is as close as I've got. Warranties are of no interest because none of those apply and very little of it needs any kind of scheduled servicing. So most of it is in my head. I've talked about in the final bullet there [referring to notes sent to interviewer], labelling of my floppy drives. Can I show you another couple of items I brought to the interview?

Interviewer: Yes please do.

Interviewee: This is an RS Systems floppy exerciser, which I bought on eBay a long time ago. It's dusty because I haven't used it for years. You basically put a power connector into there (points to Molex 4-pin socket). You connect some of these brass connectors to an oscilloscope. You then power the floppy drive with this cable so that the power signals can be picked up by the unit and you plug that onto the drive. Anyway, this thing has lots of features and I have brought in the manual that I scanned. I will send a PDF file of the manual to you.

Interviewer: Oh, please do. That's so cool. So this was used to basically test floppy disk drives. Am I getting that correct? 

Interviewee: Yes, with this kind of thing, which is a calibration disk (shows calibration disk in its box). So this one, of course, is a 3.5-inch version. And they come in two different types. One is called analogue and the other is called digital. I was very lucky. I saw somebody selling a quantity of Dysan calibration disks and made an offer which was taken up. This one (shows a 5¼ inch disk) I think has never been taken out of plastic and it is designed for 360 rpm so one would use this in a high-density drive. Here (on a paper in the box) it shows what kind of signal is written on each track. These disks were made on very specialised, highly calibrated equipment. And the idea is that you would use them with the exerciser. You would use the exerciser to be able to step to the track to pick up the particular waveform.

Interviewer: That's so cool. And again, something that I didn't know. It makes total sense that it exists because you want to be able to validate if your drive is actually working correctly. But it's so cool to see one. Very cool.

Interviewee: My view is that when you've got a known good disk the testing can be reasonably reliable without a calibration disk. That is what I was doing when I was testing these drives, I was basically saying, can I rely upon a floppy disc that I know is going to read correctly in a drive that is working properly? If I can't read the reliable floppy disk in this other drive, then there's a problem with the drive. 

Interviewer: With the drive and not with the disk which you knew. 

Interviewee: Which I might wish to address at a future stage. But you know, to date, because I've got a large collection of them and I know which ones I like, I haven't tried to re-calibrate the drives that failed to read the floppy disk.

Interviewer: It's not that, yeah.

Interviewee: So in fact there is also a calibration disk for the 8-inch drives. 

Interviewer: Oh, well, so you get all three of them. It makes sense, doesn't it? Because you want to test them on all the different drives. Very nice. So just to summarise here as well, like this is very useful, but also having actually just having a disk that you know reads well. And if that doesn't read well on a drive, you know there's something wrong with it as well. 

Interviewee: Yes. Although there are various 5¼ inch calibration disks because there are drives for 40 or 80 tracks and also calibration disks for single-sided drives.

Interviewer: Yeah. Cause that's currently our approach. So like I've got a number of disks that I know read well.

Interviewee: I think it's a perfectly valid approach. Back in the day, you would perhaps have a technician with a white coat who would know all about how to use the drive exerciser and calibration disks.

Interviewer: Yeah. 

Interviewee: This is not at all the only type that I've ever seen [pointing to RS Systems floppy exerciser] but it's the only one that I've ever had an opportunity of acquiring. And I have gone through the process of using it. Can I go on to a further point?

Interviewer: Yeah.

Interviewee: Can I pick up one of your drives?

Interviewer: Yeah, feel free to.

Interviewee: [Picks up one of the 5.25-inch floppy disk drives] This one doesn't seem to have what I thought it would have, which is the hole at the end there (points to the end of the drive near the interface where the cable connects). 

Interviewer: Why would it have a hole? I think there's another TEAC there.

Interviewee: Anyway using this I can point to what I thought I was going to find which is that very often there is a hole here [points at back of the drive] and they made a special tool to allow people to adjust the position of the stepper motor which is why both of these screws have a little bit of a fluid on them which is dried and which locks them in that position.

Interviewer: Would that be the case for all floppy disk drives?

Interviewee: No.

Interviewer: Or is this really something that TEAC did? 

Interviewee: Specific to TEAC. What you had was quite a wide wheel with a shaft on it that had a screw thread. And so making fractional turns, you could move the stepper motor position forward and back so that the drive found track zero in a floppy disk at exactly the right point. Because if track zero has been adjusted to the right place and that is picked up by a sensor here, when that (points to plastic part) passes between here, so on one side of this there is something that makes light and on here there is something that senses light and when that beam is broken by this (plastic part) then it knows it's at track zero. And trying to release these and to nudge this [stepper motor assembly] forward and back in order to get that exactly right is not something that I feel I would be able to do, except with the specialised tool made for use with the TEAC drives.

Interviewer: It seems like really specific, doesn't it? Yeah.

Interviewee: But anyway, I have seen a picture of that tool with the big metal wheel and the thin shaft with a screw thread on the end which makes contact with the stepper motor. If only somebody would manufacture those now, so that one could adjust the drives.

Interviewer: That would be lovely, wouldn't it?

Interviewee: It would.

Interviewer: What would you do if you now have a misaligned stepping motor? Would you just part it?

Interviewee: I would just leave it.

Interviewer: Yes. Because that does seem like a difficult one to fix.

Interviewee: While we're on this subject, this is a circuit board [shows picture], and I have five of them made in China which is designed to do all the things that the RS Exerciser can do. It is even cleverer in that it has various different connectors for the cables from floppy drives. So for example, that is a Shugart 50 pin connector. This is a Hewlett Packard 34 pin connector. This is for an Apple II drive connector. This is a 40 pin something or other connector. Anyway, basically these switches would operate in a similar way to the RS Exerciser. And here there are points where probes from an oscilloscope can be connected to allow you to monitor the signals and this (two-digit 7-segment display) tells you which track you're on. 

Interviewer: That's so cool. 

Interviewee: And that location is for is a chip (where again you could probably put ten of those on my little fingernail) which I would have to solder in order to make this. I've bought some of these switches from China and they would need a little bit of filing down to make them fit. But I know that the person who designed this is one of the smartest guys on the planet. He's an extraordinarily clever guy and at some stage I will build at least one of these. 

Interviewer: You should. Yeah. 

Interviewee: And if you'd like to have a bare board, if you've got colleagues here who are skilled in using a soldering iron [...].

Interviewer: I have to convince some of my colleagues, but yes. 

Interviewee: I will.

Interviewer: That is so handy, thank you.

Interviewee: I would have put one in my bag to leave with you, but I've lost track of the PCBs.

Interviewer: But that is awesome. So you've basically built a modern but like a new version of this. 

Interviewee: Well what I have is the basis of doing it and I've started to get the parts to assemble it. This is a page about the Applesauce [hands over papers], which might have been of use if you hadn't already heard about it. 

Interviewer: But it's still Applesauce such a great thing to have.

Interviewee: Both of those are for you and actually this page is what you were telling me about Nevek.

Interviewer: Yeah, because that is one of the things that I would really like to talk to you about, which is the imaging and the cleaning of disks because, so I think I mentioned in my e-mail, I'm doing some testing with our Conservation department to see what, and I know that you've done a lot of these tests as well, to see what cleaning method is best suited for floppy disks. And you probably already know, but it's one of those things where there's quite a lot of information online that you can find, but there's quite a lot of people who contradict each other. 

Interviewee: Yes. perhaps I can preface what I'm going to say. I put a message on the Vintage Computer Federation Forum which was asking 'how does one avoid the bulldozer effect'? When a floppy drive head starts to scrape oxide off a disc, it builds up more and more oxide, which then scrapes away more and more. I was fully aware that in raising this question, I was covering ground that had been covered before. But what I wanted was to draw out advice from somebody who'd come up with a new approach. The answers that I got led to cyclomethicone and they were given to me by a gentleman called Chuck Guzis who has made his living reading magnetic media for people of all different sorts. If he doesn't know something, it's not worth knowing. I was very grateful to him that he answered my question. But I would have been even happier if other people would either endorse the approach that Chuck was recommending or say, 'I've used that method but I think it's even better to do X'. So I missed out on the type of advice I was hoping to receive. At any rate, I did do what I think is good practise, based on the advice of Mr Guzis. Eventually, I added to that thread a report of mine about how it went when I used cyclomethicone. Because these forums can be terribly disjointed. Somebody starts off a thread about a topic, somebody jumps in, takes it off in a different direction, and there's an awful lot of rubbish to be waded through. 

Interviewer: I'm so happy that you did that actually because I've gone on enough forums where someone has taken someone's advice on something and then gone away and then never replied and it's like, what happened? Did it work? Did it not work? And then sometimes people asking the same question but yeah it's I think it's very good.

Interviewee: At about that time, the museum had been given a number of machines that had been the property of a UK distributor for companies in other countries. The company had kept a lot of the stuff but it had spent decades in poor storage conditions. The donation included 70 floppy disks of which most had been written on Wabash media. Now 'Wabash' that is a name that you know sets red lights flashing because while it was probably very good when it was manufactured, over time the binder that holds the oxide layer onto the mylar disk has crumbled away or become weak. And so, to explain the approach I used to prepare those disks, perhaps I can give you a picture of me standing at a sink in the utility room in my house. What I do is I get a large plastic tub, a bucket would do, and I add water a little bit hotter than hand temperature into the container, lots and lots of water. And I then put a few drops of the kind of detergent which you would use for a fine wool garment. So it is pure detergent, it hasn't got other things in it (such as the salt used in dishwashing fluids) and that goes into the container. And I let each disk soak, but I need to backtrack about how I know which disk is which. 

Interviewer: Because do you take them out of their sleeves? 

Interviewee: Yes. 

Interviewer: Yeah, you do. 

Interviewee: I cut the side of the sleeve, take out the disk medium. To start at the beginning, what I did on that occasion, was I used a small metal punch (about 2.5mm) to make a hole at a 90-degree angle to the index hole on the part of the disk that is clamped to the spindle. Given that hole, I would be confident that the underside was track zero and the top was track one. If I had flipped the disk, the hole would have appeared on the wrong side of the disk. I also have some permanent white marker pens and so I wrote my disk number onto the inner ring of the disk (on side 1). Those white marker-pen numbers survived the whole of the rest of the process that I'm going to describe. 

So I start off taking the disk out of the sleeve and I drop it into that warm water with detergent. I allow it to soak a little bit because I want any grit and other material to fall off down to the bottom. Then I lifted it out, this is what I've brought that for (shows a circular sheet of perforated silicone). Somewhere, I found this kind of silicone circle. I put it down on a flat surface. I put the wet disk on it and just rub with a thumb while the disk is wet, covered in water, just rub it at a right angle to the tracks to push any dirt off the oxide. Now I'm not sure that I ever moved any dirt but I then turn it over and do it on the other side. I think it was worth doing that. I then put it back into the water with the detergent and just kind of moved it, taking great care because the disks are very floppy when not in the sleeve. Then take it out of the first container and put it into another container which had slightly hotter water. That's because I wanted that to dissolve any of the detergent which was still on the disk. I put the disk into the rinse container and left the disk there for about twice as much time as it had been in the first container.

And the final stage of washing was to gently stick the 8-inch floppy onto a large plastic plate and just leave it standing upright in a plate rack to drip and to start drying for ten or fifteen minutes. What I then did was to tear off pieces of kitchen towel and I laid out on a worktop various squares of kitchen towel and put each of the disks on these squares when it was getting towards being dry. I left them for another 10 or 20 minutes to dry out. Then I came along with more kitchen towel and picked up any drops that were still in evidence. You know, you could just kind of let the drops be picked up by that absorbent towel. 

Then I'd put the dry disk back into its original sleeve, temporarily. The next stage was based on something that I made for myself. I had a big transparent plastic storage box. I made a 'tower' of horizontal shelves out of sheets of cardboard. The cardboard had been wrapping for flat panels of furniture. I cut the card using a sharp knife. I took two of these sections of card and I cut a series of slots in the sides of the tower. Then I took the sharp knife again and I made another set of squares, each of which I had a little end piece that would fit into those slots.  In this way, I made a tower of cardboard shelves, where again the circular pieces of silicone came into use. I used the kind of kitchen thermometer that is to be put into a joint of meat so that you can detect the temperature in the middle of the joint. The thermometer has a sensor on the end of the wire. I put that sensor on one of the upper levels of the tower. The unit displaying the temperature had a digital display which could be in Celsius or Fahrenheit. The target was to get the temperature of this box up to around about 58 degrees. But I wasn't too fussed if it was more than 55 and less than 60 degrees. In order to heat the 'oven', I had two incandescent light bulbs. These bulbs were each on a cable with a switch in it. Lastly, I had a fan that had come out of a 1990s computer case blowing across these light bulbs in order to avoid any hot spots in the oven. The fan kept the air moving. 

The box was upside down. The lid was on my workbench and then the tower was there in the middle and I could lift the container of this box over the whole apparatus with a few wires coming out of it.

Interviewer: Nice.

Interviewee: Then, if it was getting too hot, I could switch off one of the light bulbs and the idea was to leave seven or eight disks to dry out at that temperature for four hours or more. One could potentially leave it overnight, but I was a bit nervous about doing that. 

Interviewer: I can imagine.

Interviewee: So I wouldn't start a batch too late into the evening. Eventually, I worked through the whole lot of 70. After which, I made sure that the disk itself went back into its original sleeve. I wanted to maintain that link all the way through. 

Interviewer: Were you not worried that there were any like, mould or dirt on the original? 

Interviewee: Yes, I was. And so that was temporary, but I felt that it was critically important to maintain that link. Anyway, so I then came to the phase of reading and I kept these things in a sealed box where there was no air to make them moist again. And so I used the silicone sheets once again. I put one of them on a flat surface, put a few drops of the cyclomethicone on a bit of a lint-free cloth and then spread it across the disc. Once again, the direction was at right angles to the alignment of the tracks. Next, each disk was flipped  over and a small amount of cyclomethicone was applied on the other side. 

I had taken some fresh unused 8-inch disks and I had opened those sleeves. I put the disk I wanted to read into one of these new sleeves, that I knew would be clean, with the Cyclomethicone on both sides. The disc still looked wet with this surface layer. When I put it into the 8-inch drive, although the disk sleeves suggested that these were single sided disks, I decided to use a double-sided drive because if any data had gone on to side one, the upper surface, I wanted to read it in that one session. Finally, I started reading the 77 tracks. The command line sent the read/write head to something like 35 and I put on the command line 'read tracks 35 to 39'. 

Interviewer: Yeah, you were saying this so you don't read the directory.

Interviewee: Yes, but also I don't try to read the entire surface in a single go. And then when it stopped after having read track 39, I would take the disk out of the drive and look carefully at the read-write head. And once again, I haven't brought in what I intended to bring, but of course I could just have described it. This is a coffee stirrer. I take a cotton bud and I cut it in half and I use bit of Sellotape to fix it onto the end of it so I can dip it into a pot of IPA (isopropyl alcohol) and I can then reach from the side of the drive and rub the read-write head across this way because that is avoiding putting any scratches on it. If one did it from the front of the drive there would be a risk of making the surface less smooth and at right angles to the movement of the disk. What I would do with my left hand would be to put gentle pressure on the top head while I was moving the cotton bud saturated in IPA from left to right until such time as I felt that the read-write head was entirely white and there was no evidence of brown oxide left on it. That was often really difficult to do.

On a few occasions, I did resort to using a heavier stick and I put a bit of cotton bandage around it. Again, this was a non-abrasive surface which would hold even more isopropyl alcohol, but the stick was slightly more rigid. Most of the effort needed to be on the lower head because it's the bottom head that tends to pick up most of the oxide. I could put a bit of pressure near the lower read-write head with a stronger stick, in order to get rid of all of the oxide that had come off. Because that is crucial because if oxide is left, it will just collect more oxide from the next tracks on the disk.

Interviewer: It would do the same, wouldn't it? Like it would just go and scrape. 

Interviewee: So it is a lengthy process end to end, from the putting the label with the disk number on it, originally photographing it, then doing the washing, then the drying, and then the reading. 

Interviewer: And then, yeah, because that's also really time intensive, yeah. 

Interviewee: There's a point I should emphasise, which I haven't mentioned so far. When reading a disk, it's important that the drive is placed on a work surface, not encased in a box or in the computer chassis. The ability to check whether oxide has started to build up on the read/write heads, and to be able to access the heads to clean them, depends on having a clear view of the heads. It helps to have a bright light to illuminate that area. This applies also to 3.5 inch drives. The top of the case can be removed. I'm generally not in favour of the use of 'cleaning disks' which I think can be abrasive, even when a few spots of isopropyl alcohol have been added. They were good enough when the disks were new and fresh, but I prefer to access the heads directly and to make use of plenty of isopropyl alcohol.

So, returning to the reading of those 'Wabash' disks, I believe that all 70 of those discs would be able to be read again using a similar process. So I was not trying to get every single sector read on that occasion. What I wanted to be able to do was to say this disc has got that kind of data on it. And actually much of the software was quite conventional popular software in packages like Word, WordStar, SuperCalc and D-Base II, where there was no particular intrinsic value to have the one that had been written for the floppy disk format that the particular computer system used. So in a sense, the proprietary or specialised data which was specific to that computer system was only present, at least in significant quantities, on a small subset of those 70 disks.  In fact, some of the disks turned out to belong to a different computer system.

Interviewer: Oh, that's interesting too. 

Interviewee: There were clues on the labels. 

Interviewer: The labels gave you clues. 

Interviewee: But the labels only gave a clue. 

Interviewer: You know, could be certain. You never can be certain. 

Interviewee: Do you know that with an 8-inch disk, the index hole is in one position if it's single sided and in a different position if it's double-sided? On a single-sided disk, the index hole is at about 7 degrees to the right of a vertical line. On a double-sided disk, the index hole is at about 26 degrees to the right. An 8-inch drive with upper and lower read/write heads usually has two sensors, one of which is for each position of the index holes. Whether the encoding is single-density or double-density is a separate issue.

Interviewer: Yes. But that does sound like, so would you only, because it's a time-intensive process to do it that way. Would you only recommend doing that on discs that have the oxide that is starting to scrape off? Would you say...

Interviewee: I think when one is given a collection, say 10 or 15 disks, and it is not a matter of life or death, but you really want to be able to read these disks [...]. What I look for is a backup or a duplicate. And of course, I take each disk out of the [paper] envelope in which it is stored and look at the oxide top and bottom and then turn it round by just turning it with fingers to notice up any scratches or mould or other defects. But I then start by reading a few tracks at around the middle track. It is possible to view that small number of tracks with  software like HxC floppy emulator. It's a really useful piece of software. The partial disk image can be put it in the dummy disk view which shows you a picture of the disk, with side 0 and side 1 both in view on the display. You can see how many sectors there are. You can see very often that the outer tracks have read nicely (shown in green) but the inner tracks may have red spots where there have been problems in reading part of a sector. On the inner tracks, the sectors are packed more tightly. This view of the disk allows you to get a sense about how well the rest of the reading is likely to go. If there is a lot of red on the view of the disk, this could mean the drive of not the correct type for the disk, for example a high density disk in a low density drive. If there is a patch of red, this can mean the disk was at some time affected by a magnetic field.

Of course most of the time everything goes well. Rather than reading only a small number of tracks at a time, particularly when you've read four or five disks with no oxide being left on the heads, one can start extending the range of tracks. I advocate always starting with a small number of tracks, avoiding the location of the directory, for each disk. This is because the risk of oxide coming off is specific to the disk; it's not a property that all the disks in the set are bound to share. 

Interviewer: Which is also interesting because it's also sometimes specific manufacturers that have that problem. 

Interviewee: One of the sets of disks that I started to read had about 150 Newbrain disks. These disks were donated to the museum in a supermarket's plastic bag. I presume they had been stored for years in a garage or a garden shed ('garden shed floppies'). After I'd read about half this set, I decided to stop at this point. I found I was running into trouble too many times. This was long before I knew about cyclomethicone. It didn't seem there was any fascinating data to be gathered from these disks, judging by what I had seen up to that point. It just seemed better to leave the second half. This is a little bit like archaeology, where a team will only dig part of a site because in 20 years' time there may be more developed methods.

Interviewer: Somebody may have better techniques, yeah. Oh, fascinating. I like your whole term as well, garden shed floppies. There's a number of people in our community that have done, that have found floppies in garages or lofts, that type of places, and ran into similar problems.

Interviewee: Basically, disks that have been kept in these [original] cardboard boxes are likely to be good because they'll be clean at least, even if the temperatures in which they had been stored had gone up and down.

Interviewer: But it seems the carrier bag; it's not the recommended way. I especially really like your approach of going for a small number of tracks in the middle of the disk, getting an idea of what's actually on there and if it's actually of importance. 

Interviewee: Yes, I would tend to do that as a later stage because really what I'm trying to do at the beginning is just to you know to see what are the chances of getting a good read of the disks.

Interviewer: Yeah, so that's when you use [the HxC] floppy emulator and have a look at the actual tracks in the sectors and see how clean it is. Yeah so that's like approach number one.

Interviewee: And part of what I felt my role was, was to do that initial check of a disk. The log file, which I very often kept, was intended to allow anybody who wanted to make more serious use of that disk to begin with information about which tracks had shown issues on the occasion when I had read the disk. The log file made with the Catweasel Tools software recorded how many attempts had been made previously to read the sectors of each track. They would be able to look at the disk surface, presumably under a magnifying glass or microscope, to see whether there was an area in that region of the disc that showed surface damage or whatever. 

Interviewer: Conservation's got a microscope downstairs and I'm super excited.

Interviewee: I actually bought one on the basis that the grandchildren could look at bugs. 

Interviewer: Oh, there you go. That's a great way.

Interviewee: I have it on my workbench. If I want to look at something closely, it gives an amazingly clear image. Its cost was a few hundred pounds. 

Interviewer: Yeah, it's so cool, isn't it? 

Interviewee: It's been one of these digital things, but it has proved its worth. 

Interviewer: Yeah, which is good to hear because microscopes seem to be a very useful thing to use. Shall we move on to discussing some tools and software?

Interviewee: Which I think is on the second page [refers to notes sent before interview]. 

Interviewer: Yes. Because I think that, well, you've imaged a disk, you've got like either a flux stream. So our approach here is I will always make a flux stream, but I will also if the formatting is clear, I will also get it in that specific format. So like that's the two things that we get. Is there anything? What's your approach?

Interviewee: A brief comment is that flux stream images take up a lot of space. I almost always use 7z or zip (winzip) to compress them. What I haven't talked about is my serial numbering approach. At a very early stage, I adopted a convention about numbering the disks and disk images. For example, if it was an 8-inch Intel disk, I would use 'INTL' as a prefix, because 'INT' is far too common. I would normally aim for three or four alphabetic characters which identify the manufacture of the computer system. That code was followed by some zeros and allowed counting up from 001 to 999. What I did later on was to put either 3, 5 or 8 in the middle before those zeros to identify the size of the disk.

When I started, I would use a fine tipped pen to write the resulting disk ID on one of the labels, normally the user label. A further standard procedure was to photograph every disk. When I've photographed a batch of disks, I have used a DOS batch file to convert the image number that the camera or my phone gave. I've read out the list of photo numbers into a spreadsheet and then I've replaced it with my disk serial number. Coming back to the example of Intel disks, I would copy the prefix 'INTL' into a column of cells. In the next column, I would have the three, five or eight. Then in the third column, I would enter a string value of zeroes with an initial apostrophe. In the first cell of the column, I would enter 1 on the top cell. In the next cell I would enter the formula (cell column and row values), plus 1. Then I'd copy that formula down as many cells as required. Because that remains as an 'active' formula I'd copy the whole list of numbers. They might run from, say, one to 125. I'd copy that into another column ('copy as value') as it appears in text so it loses its property of being a formula. Finally, I would over-write the formula column with the text string, running from '001' to '125'. Next, I would concatenate all the elements of the disk number into another column which I would then use to print out a sheet of labels. This text value could then have a suffix such as '.dmk' or 'imd' or 'jpg' according to whether it was a disk image or the photograph of the disk. The aim of this process was to ensure consistency in the identification of the floppy disk itself, the disk image or images and the photograph of the disk in the paper sleeve.

Interviewer: And then you would use that to number?

Interviewee: I would attach the small adhesive labels (about 10mm by 15mm, with 270 blank labels on an A4 sheet) to the disk before photographing the disk and the paper envelope. I normally aimed to attach the disk number in the top right on the 'user label' if that area was free. Or in the bottom right if the top right had text. If that area also wasn't free, I would put the label  on to the manufacturer label, because knowing the manufacturer of the label didn't seem as crucial as adding a clear reference number that was visible when the disk was in the paper envelope.

Interviewer: Yeah, it's not that of interest. Because you list quite a bunch of software tools here that you do use [referring to notes sent by Interviewee before interview]. And also interesting to see that the use of vintage or older machines as well. Could you elaborate a little bit on that? 

Interviewee: I'm comfortable using command line software, particularly when you can just press an up arrow and recover the whole command line from the previous command. Then one can increment a couple of digits and that's what you want for making an image of the next disk in the set. And again, the structure of the command line is a property of the Catweasel Tools, which is the software that works with the Catweasel.

At a later stage, for example with a CP/M disk, there's a software utility called '22Disk', which allows the files to be written to a hard drive in a format that works with MS-DOS. '22Disk' was originally written by the same person, Chuck Guzis, whose name I mentioned earlier. This program works with a file of 'disk definitions' that is able to handle about a couple of hundred CP/M disk formats. These disk definitions are saved in a text file. Additional definitions can be added to the list, which then needs to be processed by a utility that is also part of the software package. There have been several individuals who have explained how an additional disk definition can be worked out. An alternative means of exporting files from a CP/M floppy image is called 'CPMTools'. This is available in versions for Linux and MS-DOS. These require a format definition file with a similar structure to that used by 22Disk and a similar number of 'ready made' definitions.

Interviewer: That's a lot.

Interviewee: With 22Disk, the deal was that if you hadn't bought a licence, you would have access to definitions for something like 40 or 50 common formats. But if you wanted the full list of definitions you had to pay for a licence which was something like $30. Chuck Guzis is no longer the licence holder. He's sold the program to another company. 

There's a man called Larry Kraemer, whose name I've mentioned on the last page [referring to notes sent before interview], who is the person to go to if you're faced with a CPM format and you're having difficulty writing a code to be able to read it with either 22Disk or CPM tools. 

CPM tools, by its essential nature, is a piece of software for use with Linux. Under Linux there is a package called LibDSK written by John Elliott. This software is described as a 'library' for accessing floppy disks and disk image files.  It can be used to provide support in various situations. These situations include the use of emulators of a particular computer. LibDSK can also be used alongside CPMTools.

The person, who designed the Greaseweazle (floppy disk hardware and software) and the FlashFloppy software (for the Gotek floppy emulator) (named Keir Fraser), also wrote a software utility called 'Disk Utilities'. Similar to LibDSK, this provides support for making and manipulating disk images, including files capturing flux transitions. It is provided as source code files and needs to be compiled to an executable that can be run under Linux, Windows and MacOS.
 
Interviewer: I'm interviewing him [Keir Fraser] on Friday, which is exciting.

Interviewee: He's such a good bloke.

Interviewer: He's so nice. 

Interviewee: Under Linux, CPMTools is able to make use of the LibDSK 'library' and that just makes it work better reading out the files or providing a list of the files on a disk. 

I also use a programme called EditDisc. This works with hard drive images as well as floppy disk images in MS-DOS format. It provides a convenient way of extracting a file-system from hard drives, preserving the hierarchy of directories and files. 

Interviewer: Oh that's nice actually. 

Interviewee: And there is another package which will run on a modern computer which is called DOSBox. 

Interviewer: Yes, DOSBox is one that I have heard of. But yeah all these are all new to me. 

Interviewee: And I put in brackets there [on notes provided in advance of the interview] about the use of 22Disk definitions. On occasions when I know the 22Disc specification for the floppy disks, (which includes the number of  tracks, how many sectors per track, where track number where the directory is written, how many directory entries were available, and all of that), I have put an IMG file, which is just sectors without any formatting data, into a Gotek with Keir Fraser's FlashFloppy software. And, using 22Disk, I've been able to read out the files from a disk image file to an MS-DOS directory, without needing to read the physical floppy disk again. 

Interviewer: That's useful.

Interviewee: And ImageDisk is also mentioned on the notes I provided. This is a program for MS-DOS. It was written by an American called Dave Dunfield. Like Chuck Guzis, Mr Dunfield is now elderly. He wrote a brilliant software package consisting of ImageDisk and its several associated tools. One of those tools is called 'DMK2IMD.com'. This is a utility I use with MS-DOS batch files to take a whole set of DMK files that I've produced with disk images processed on the Catweasel device. In a few seconds, by running a batch file, I can have a matching set of IMD files. 

Interviewer: That's really cool. 

Interviewee: And there was one occasion when an IMG disk image needed a special format with track zero, side zero being FM, single density, all the rest of the image double density. There is a tool called 'BIN2IMD', also part of the ImageDisk package. This allows the user to write a physical copy of a disk with different encoding on different tracks. With BIN2IMD, one can use an 'IMG' file (which consists of data organised as tracks and sectors), but where the IMG file doesn't specify the format of the physical disk. The utility creates a disk image for use with ImageDisk, to which has been added the information needed for the disk to be written with each track and side having the required encoding. It is important to mention that ImageDisk doesn't require any specialised hardware. It is used with an MS-DOS computer (but it doesn't work well in a Windows 'DOS Window').

When I mentioned the idea of my setting up some of my machines at the museum and actually showing you how I do these things, which in some ways would be better than just talking about them, that is the kind of workflow that I'm thinking about. 

Interviewer: That is awesome. I will have to go and explore all those tools. But again, like I think the community will be really interested in hearing about all the different pieces of software that are out there. Because that's another thing that can be really difficult to find. You kind of need to know someone who knows about them, if that makes sense. Which can make it really, difficult.

Interviewee: There are many other disk imaging packages which have features making them suited to working with disk images for various computer systems. There's a guy called Simon Owen who produced a package called SamDisk, which is really good with BBC Micro disks and stuff of that sort. There are just lots and lots of disk images packages. I'm sure that people develop trust in their personal favourites.

Interviewer: And I think that's as well like a thing, isn't it? It's like for us as a community, it's more to know what ones can do what disk types. 

Interviewee: I don't think anybody has written the encyclopaedia of disk images.

Interviewer: Because there are so many of those tools out there and even ones that are no longer updated can be really useful sometimes. 

Interviewee: There is a later version of the ImageDisk utility programs (not the full-screen 'ImageDisk' program itself)  where somebody else found a bug in what Dave Dunfield wrote and they've remedied it. 

This is Mark Ogden. He has provided support for my efforts on many occasions and has made available the tools he developed to help me. Among other things, he has provided 32-bit versions of Dave Dunfield's 16-bit software. 

Interviewer: Oh cool.

Interviewee: Sometimes, you know, because I'm using a 64-bit machine most of the time, I may try to run one of these 32-bit versions and it doesn't work. But it's easier to use a different machine that does have MS-DOS or an old version of Windows on it.

Interviewer: That is another reason to have an older version, isn't it? 

Interviewee: Towards the bottom of page two [refer to notes sent before interview], I've listed tools to manage an archive of disk images. I've mentioned that I am aware that in the community there is very specialised software for digital repositories, such as Archivematica. I have looked into others as well. None of those approaches were appropriate for a one-man band. 

There are three Windows software packages that I found tremendously useful. TreeSize, which I mentioned earlier, is very powerful. I buy the licences three years at a time. It's 40 or 50 euros for a licence. It is something I use numerous times every week in managing data. As the name implies, it displays the directories on a hard drive (or a floppy drive) as a 'tree' structure, with the storage space occupied and the number of sub-directories and the number of files. It also has tools for finding duplicate files. Very recently, I realised that it has added the feature of locating duplicate directories, based on the large number of users who had asked for that as a feature. It allows duplicate directories to be identified on multiple hard drives.

A second useful program is WinCatalog. It is written by somebody from Eastern Europe. You can put optical disks, hard drives and so forth, flash drives, virtually anything you can connect to your computer. WinCatalog will read each storage device and builds up a database of files. A feature I use frequently is to search for a specific file in the database, before then accessing the hard drive or other storage medium to access the file. The database files can be updated to reflect changes in the directories and files on a specific storage device.

Interviewer: Of course. 

Interviewee: So, I've got WinCatalog database files which are as big as five gigabytes, which hold records of the files I've stored on several large capacity hard drives. These hard drives are backups, in what I would call cold storage. They're not actively spinning the whole time. But I go to the catalogue that I made using this software, I find exactly where I need to look. Often I know exactly which file I'm looking for. When you've got all that material in one database, you can do a search across all of it. It works best if you can remember the file name you're looking for or the directory name.  But it is also possible to specify a text string that occurs in a file and to list the files containing that string.

Interviewer: Which is really useful, isn't it? 

Interviewee: And the last one of the tools I have found useful, FreeFileSync. For as long as I've used computers, I've been keeping backups. The result is that I've got multiple versions of a directory that has, to give an example, Intel disk images on it. The backups include one I saved five years ago, another that I saved last year. Using the FreeFileSync software, I can identify a backup file which I think is the better copy of the data. Then, I can identify another backup file that has some corresponding files, but fewer files because it was saved at an earlier date. The software generates a list of all the files on both of the directories or disks. The list identifies the files that are not currently on the target directory. And it will just copy those files to make sure that the target directory now contains all the files in a certain collection. 

Interviewer: Just the ones that have already there.

Interviewee: Without overwriting the files that are common to both of them. 

Interviewer: Oh, that's really useful. 

Interviewee: And it's called FreeFileSync because you don't have to pay money for it, although you are invited to contribute to it. That software package plays a useful role in managing a large collection of data at a later stage. I have found it reliable, even when it's copying many gigabytes of data from one directory to another. It just seems to be reliable. 

Interviewer: Which is really good, isn't it? Once you find a piece of software that works, it's just nice to have it to work, isn't it? 

Interviewee: Yes, I agree.

Interviewer: And then I'm just going through the other questions. So we had a chat about trying to read on a different drive.

Interviewee: So, floppy control app. That's the name of this thing that I showed you with the micro stepper.

So this is where I talked about this Catweasel tools software which works out the appropriate format of a disk whose format may be unknown at the start. 

Interviewer: Does that only work with the Catweasel? You need the Catweasel hardware, right? You need the floppy controller. So that's not going to be... Well hopefully. I'll keep my hopes up that one day I will find a Catweasel. 

Interviewee: Yes, well I have a permanent search on eBay.

Interviewer: Me too.

Interviewee: There is a kind of fluffy toy that comes up. 

Interviewer: Yes. Oh, I know exactly what you mean. Which is annoying. Well I was wondering about as well and I will maybe ask [Keir Fraser] about this on Friday. Do you know if there's any opportunity for the Greaseweazle to do similar to the Catweasel where in some ways it can recognise what's there?

Interviewee: I think you'll have a great opportunity to ask Keir.

Interviewer: I will ask him.

Interviewee: I think that he is making the Greaseweazle software better, that's one of the great things about him. He's working on it and increasing its scope over time. I think he might well be prepared to include features that have appeared in other disk imaging software.

Interviewer: I would hope so. I will ask him because it seems like something that may be feasible, but yeah.

Interviewee: What it is basically is reading the flux transitions or a specified disk format and working out, "Does this make sense? Perhaps better than putting it that way 'does this match a conventional disk format and encoding'? What are the common encoding arrangements? And if it has already read track zero, side zero, when it steps from track zero to what it thinks is going to be track one, if it finds there's no data on track one, then it's got to double step for the whole of the disk. It's quite a logical process. In that situation, the floppy disk has 40 tracks of data (at a spacing of 48 tracks per inch), but it is being read in a disk drive that is able to read and write 80 tracks (at a spacing of 96 tracks per inch). The combinations of a 40-track floppy in an 80-track drive will often produce a good disk image. However, if the flux transitions are harder to record, then it will generally be better to read that 40-track disk in a drive that is designed for 40 track disks. The reason for that has to do with the better match between the width of the read head and the width of the waveform data laid down in the oxide of the disk.

Interviewer: Yeah, it would be so nice, wouldn't it?

Interviewee: And if [Keir Fraser] wanted to know how the double-step feature had been implemented for the Catweasel Tools software, the source code of that software is available and he can see how it was done.

Interviewer: Yeah, which makes it a lot, because that is the thing that sometimes I kind of worry about a little bit with the Greaseweazle, because sometimes you make like a first image of something and then you're like, "Oh, okay, I'm looking at this format." And of course, you can go from a flux stream and you can convert that. But it's always something where I'm like, it would be so nice.

Interviewee: In a sense, you've got to know quite a lot about a disk in order to get a valid flux stream. It would take you further forward if you had a system that was able to detect that sort of characteristic of the disk on the basis of the data being read from a few tracks.

Interviewer: Detect it. Yeah, I will mention that on Friday to [Keir Fraser]. 

Interviewee: One of the resources I'd like to point to towards is a book called Forensic Computing by Sammes and Jenkinson [ISBN 1-84628-397-3]. When I say that, don't buy it new. I bought a new copy of the hardback edition for something like seven pounds because they clearly printed more copies than there were people who wanted to buy them. And so, on Amazon, there are often used copies at something like five pounds. 

Interviewer: Well, I have to get one of those.

Interviewee: The book is all about forensic computing. It covers topics such as 'where could a criminal hide useful data in the crevices of a hard drive'. A typical hard drive doesn't use literally every square millimetre of disc surface. There are hiding places. That book goes through the fundamentals of how magnetic flux transitions are recorded and so forth. It goes deeply into these topics. 

Another point the book emphasises is that in forensic applications it is vital to ensure the integrity of the original image of a disk, whether that's a floppy disk or a hard drive. I think it's likely in the University Library's handling of digital material that there is a similar emphasis on the integrity of the digital document. The book may have some pointers to recommended practices.

Interviewer: Which is exactly what we want. That's really cool. 

Interviewee: It's useful information. 

By the way, I've also mentioned Phil Pemberton [in the notes responding to the question topics]. He did a student project and designed a device called a 'Disk Ferret'. I have asked him to let me have a printed circuit board for the Disk Ferret, which I would donate to the museum. However, it needs some advanced soldering skills to make a working device. Anyway, Phil Pemberton is someone who has experience of using the idea of reading a track from a floppy disk into the memory of an oscilloscope.  The idea is to record that data as a waveform, so that you can then look at the strength of the signals and edit the waveform to edit flux transitions that were not read accurately. He and a friend of his used audio editing software to do that with success and it's all written up in a blog post ['Hacking Everything', Blog: scarybeasts, by Chris Evans, May 17, 2021]. 

Interviewer: Oh, awesome, I have to go and look at that. 

Interviewee: I have a directory which I call ‘floppy imaging tools’. It contains a library of material there that goes back years that I've been gathering this stuff.

Interviewer: But that's also so important because there's so much of that stuff that's just like not out there anymore because websites have not been archived or like manuals. They're kind of difficult sometimes. People just throw them out because they're manuals. Because currently we're cleaning out one of the cupboards downstairs and there's a whole pile of old manuals for like, there's a Linux operating system and there's some like other. And I'm nearly at that point where I'm like, "No, we shouldn't throw them away. What if like in 40 years' time someone needs them? It's a worry, isn't it? 

Interviewee: A final point I'd like to mention is that the Cambridge Computer History Museum has two 'format conversion' machines that were built in the 1980s. These computers were able, for example, to take a 3-inch disk, read it onto the hard drive and write it out on a 5.25-inch disk. One of them came from a local government IT department. I guess that in the late 80s they had some people using Amstrad, some people using Apricot, some on BBCs, people working at home using other things. The 'disk conversion' machine and its several drive types and the specially written BIOS software overcame the problem of incompatible disk formats which existed at that time.

Interviewer: Yeah.

Interviewee: That machine could read numerous CP/M disk formats and some others. It has a library of 700 formats. When I analysed the data files I found there were about 150 unique formats. It was obviously easier for a user to pick the name of the computer on which a disk had been formatted and written. I met somebody who often comes to the museum when there's an event like the 'Retro Computer Festival'. He had bought another of the disk conversion machines for his own collection. By chance he brought all the printout that he had received from the original owner of that machine.  I was able to borrow the documentation, took it home, scanned it all, gave it back to him the next day. Each page of the many pages that I scanned has a detailed specification of a particular format. 

Interviewer: Wow! Were there formats that you didn't know?

Interviewee: I have not really put it to use. However, I've checked with the owner of that documentation. He's happy for me to pass the scanned version of the data to other people.

Interviewer: Oh, please. I would love a copy of that. 

Interviewee: It sounds to me that there are quite a number of things I should put onto a flash drive with a fairly large capacity and let you write those files on to one of your machines. 

Interviewer: Yes, please. Do let me know if you need a flash drive, because I'm also happy to send you one and then doing it that way if that's easier for you. Because yeah, I would love anything. That sounds very useful. And I do want to stay in contact because this has been really useful. But also just in general, thank you for the interview.

[END]