1. I have been having something of a rapprochement with cement splicing, with which I have in the past had little truck. I am not really sure quite why. Maybe partly because the CIR-type tape splicer, designed with magnetic stripe in mind, does not fully cover both sides of the join, which feature is intended to leave the stripe uncovered. This is not really relevant on 9.5, and I feel it gives a slightly temporary air to a join, but it can readily be overcome. Although tape does not, in my experience, dry out if it is in the body of the film, it might do so if it is near the front end. The front is very prone to drying anyway, and this often causes problems with leader and titles becoming brittle and cracking and, indeed, with cement splices, which warp, especially if made in the traditional manner with far too much cement. Whatever the reason, I have been using cement increasingly for permanent attachment of leader in particular. However, Chris Bird has made the excellent point that, even if tape does dry out at the front of a film, it is easy to replace and no frames are lost as they would be with cement splices. It would therefore be better to use tape than cement at the start. I think he's right and I will henceforth follow his advice.
2. Tape is, of course, unmatched in its ability to repair, and I have saved films that would otherwise have lost valuable frames of soundtrack, or would have broken on projection or been completely unprojectable. More on this below.
3. To revert to cement. I have had to learn how to do cement splicing in a way that seems to me correct – I am open to debate on this – as it is something with which I have had little success in the past. It is painfully and appallingly obvious that I have not been alone in this problem if you look at much 9.5.
4. It seems to me that there are two fairly fundamental problems with cement splicing:-

a). We do not have professional quality splicers. This means that they are unlikely to be aligned with any great accuracy or repeatability, or to scrape or cut the film terribly well or  precisely, are impossible to adjust, make oversize splices, allow cement to leak from the join, allow film to slip and move and etc and etc. You only have to look at those beautiful, clean, narrow splices you sometimes see at reel joins in 900 footers to see what a professional splice is like. We can’t hope to match that, but we can do a lot better – I think I really mean infinitely better – than at least 95% of all 9.5 cement splices I have ever seen, and I have seen thousands.

b). The film itself is completely inconsistent. Bear in mind it was mostly made from 35mm slit into three. Width can vary, as can the precise location of the sprocket holes in relation to the edge of the film. There is also quite a bit of 16mm pitch stock around, or leader perforated from 16. The difference between the pitch of 9.5 and that of 16 is small (to be precise, 16mm is 0.3" or 7.62mm and 9.5 is 7.54mm), but it becomes significant when joining film in a splicer. There is also, of course, the problem of shrinkage, both as between different films, between film and leader and even within the same reel of film. A splicer is not usually adjustable, and has to pick some arbitrary value between the different realities. If a film is too badly shrunken, it can be impossible to get it on the splicer without causing damage by stretching the film over the pins.

5. There are many different cement splicers and they don’t all make a splice in the same way. Where the cut is made and the overlap and amount of emulsion to be scraped off can vary. This can be a problem when re-making a dried-out splice without cutting out any more frames. My current favourite is the Muray shown here, tho’ it does have its faults and idiosyncrasies – the springing of the upper leaves is far too strong, for one thing. I like the fact that it makes a fairly narrow join, and I like the way it provides for scraping the emulsion on one side of the join and the base on the other side. This extra scraping will, I hope, lead to better adhesion and a slightly thinner and therefore less obtrusive splice.
6. One curiosity with slightly shrunken film is that it seems to matter which of the two pins that will hold it I put a piece of film on first. One way round, the film definitely goes on easier and with less straining of the sprockets.
7. One of the biggest problems I (and from observation many others) have had with splices is getting them aligned properly so there are no protruding bits at the side of the film to catch on spools or gates etc. I can think I have it lined up, but when I open up the splicer to remove the film, the splice may be way out of line. I have had to resort to placing both ends of the join on the lower plates and manually aligning them before closing the upper leaves to hold the film. (This is sometimes impossible to get quite right because of differing lateral placement of sprocket holes in different stock, eg when attaching leader.) Even then, I was getting it wrong and in the end I was forced to examine the alignment thru a projector lens to get it right. This may, however, be due to my astigmatism and so not affect others. Despite all my care, I still often have to lightly trim the edges of the film at the splice to ensure a smooth join. I keep a sharp craft knife to hand for this. This does not mean, however, that you can correct gross misalignment in this way.
8. The other big issue, of course, is the application of cement. I have settled on using Agfa Cinecol, mostly because it has a brush that I feel comfortable with – each to their own on this, I suspect. And I have painfully learned that the only way to avoid a horrible messy splice seems to be to use very, very little cement. I scrape the brush on the inside of the neck of the bottle half a dozen times, which also flattens it out a bit. I try not to touch the edge of the upper leaf of the splicer on the scraped side of the join, because then the cement seems to flow under the splice by some sort of capillary action and mark the base side near the join. So I simply place the brush on one side of the sprocket hole and draw it outwards rather than along, repeat for the other side and maybe a quick dab in the centre. I try to avoid brushing across the joint or making repeated contact with the film as this seems to lead to less than satisfactory splices.
9. One thing I do, almost out of faith, is to press the joint together as hard as I can for about 10 seconds after making it. I do this in the not-terribly-well-substantiated belief that this helps to press air out and ensure full film/cement/film contact right across the join.
10. What I get when all goes well is a reasonably narrow splice, with no air gaps showing where there is no cement, and no bare area from having scraped too much emulsion off. Even now, I am still doubtful that I have actually made a good and durable splice – surely it needs more cement! I would like to hear other people’s views, experience, favoured splicer etc.
11. Because I am concerned to preserve films as much as possible, I try to re-join rather than re-make existing splices wherever possible and so save two full frames. A splice may have substantial dried out areas and be obviously heading for separation in the near future but otherwise reasonable. I find that inveigling some more cement into the join can work and avoid any further cuts. As long as some of the cement is still attached, there are no problems of alignment and the gap is usually so narrow that capillary action draws the cement in. I rely on the fact that cement does not take on the emulsion side and if at all possible work from this side. I’m afraid I simply use my fingers to squeeze the film together – as long as one avoids getting any cement on the base side, or on those fingers that will touch the base, it is usually possible to avoid leaving cement fingerprints and greasy marks can just be cleaned off. Sometimes the only way to get cement in is from the base side, but even here accuracy and speed can avoid the worst effects. You just need to practice a lot. Sometimes I just tape over a dodgy cement splice if I think it will not re-cement properly – I have felt much more comfortable about this sort of thing since reading in a Kodak film care manual that overlapped tape joins are perfectly acceptable in 35mm practice.

12. Here is a pic of a 9.5 CIR splicer. First point is that I am not terribly happy with the tape-cutting performance of any of the CIR splicers I have tried; too often, the edge is ragged. If I am doing much tape work, I tend to use a craft knife instead to cut the tape at the edges of the film. I am considering removing the blade altogether.
13.  I mentioned above that the CIR splicer does not cover both sides of the film fully. I overcome this by using a craft knife to cut the tape conventionally at one side, as the built-in knife would do, but then trim along the edge of the film on the bed of the splicer rather than leaving the wrap-round piece on. Turn the film and repeat. (This is basically what the 16 and 35 versions do – no wrap-round.) It is important to remove the wrap- round piece from the bed of the splicer, as it otherwise gets picked up by the next normal tape splice (or repair) you make. As I have said before, I find an overlap, however slight, to be essential – floppy joins are a killer on 9.5.
14. I use the same tape for 9.5 as for 8 – it’s only about 9mm wide. This width allows you to reinforce two damaged sprockets at the same time. It covers rather more than a full frame and I think is less obtrusive – it’s the edge of the tape that shows up most on the screen, and this way there is only a single line for two frames repaired. The tape can be applied on the left or right; if left, a craft knife or similar blade is essential as the built-in knife will not cut the tape properly.

15. You need to get the tape up against the pin, but not so as to curve round it, as it will then end up partly covering the sprocket hole. There may be uses for a different size of tape; on 17.5 I use 9mm and 15mm depending on the specific requirement.
16. Edge splits, rough patches, nicks etc can easily be repaired by drawing the tape only a short distance across the film, then making a freehand cut with a craft knife at the rear to provide a similar-sized wrap-round. This is one area where a wrap really is needed. In cases like this, where the leading edge of the tape will actually be used, fingers are going to leave marks. I use a very narrow screwdriver to catch the sticky side of the tape, with a finger on top for grip. This still leaves a mark on the tape but not as bad as the mark and grease of fingers. I have found that the tape originally supplied with the CIR marked permanently. I use PEC tape; marks usually vanish when the tape is pressed down. By the by, I really cannot do any of this intricate work in gloves, which get bits on the tape anyway.
17. As with cement splicing, sometimes film has shrunk so much that to get it even over the pins either of the splicing point can damage the sprocket holes, let alone the outer pins as well. This, of course, applies only in repair work, not when you have cut the film for a splice. I find I have to fudge – I fit the film to only one pin and get the tape on as best I can. I then lower the film presser strip (that thing hinged at the same place as the punch head) but insert the tip of my finger to prevent it going all the way down onto the film and so forcing the sprocket holes over any of the other pins. I then carefully operate the punch head to pierce the tape without pressing the film itself down onto the other pins.