Bookhouse’s software for finding and fixing stacking

Ah, stacking. It’s a distracting little critter. For the reader, it has the potential to create a confusing reading experience.

Regardless of from what perspective you encounter them – audience or editor – they are undesirable.

Interestingly, as typesetters we have noticed that stacking will occur more in some books than others. It can be readily associated with the style of writing or subject, but it is unpredictable. Unsurprisingly, ‘the’ is the most common stacking we encounter.

As typesetters, we are always looking out for them. If we miss them, the proofreader will mark them up, but frequently the composition of the paragraph will be altered by some other correction, so the instances they observe may not exist after corrections are taken in. Other instances will be revealed.

There is no easy way for an editor to find these – Adobe Acrobat does not have a ‘Highlight stacking’ function. A visual scan is the only tool available.

At Bookhouse, we have been working on a robust solution for a few years. Our first attempt took a very literal approach to the problem. It was a simple line-by-line parser which asked, ‘Is the end of this line the same as the previous one?’ When it encountered ‘yes!’, it would bring the page into focus and hand the problem back to the typesetter. At the time, this was a major step forward from simply scanning the page by eye – some are better at that than others. It certainly did not replace our usual checks, but it enabled us to do a ‘polish’ on the typesetting that otherwise would have taken hours, perhaps to find only a single instance.

The abilities of that software slowly progressed. The first major step was to scan the start of the line as well as the end. From a software perspective, that was not as straightforward as it sounds. It was definitely a welcome enhancement as it made a difference to how thorough we could be.

After a bit of analysis, we realised that there was actually a limited number of ways to deal with stacking, once we found it. Fundamentally, the top word in the stacking pair had to stay with its previous word while the underdog had to stay with its subsequent word. Without that constraint, the words could always end up stacked no matter what solution we attempted. So we enhanced the software by automating that step. This made a difference to how quickly we could apply the polish. A small step, but a useful one.

Incrementally, our tireless stacking assistant has been imbued with additional abilities. We analysed all the ways the problem of stacking could be solved and incrementally automated each solution. The software would make one change at a time, automatically, pausing to ask the typesetter if the solution was good or bad after each change. It was somewhat clunky, but we were glad to have a tool that could quickly iterate through multiple solutions allowing us to select the one which looked best.

The most recent update to our stacking utility is able to determine if a solution has impacted the paragraph in a negative way, by looking at the differences in spacing between the affected lines. If the spacing has not altered beyond the limits set by the typography settings – and there has been no change in the number of lines in the paragraph – then it will move on to the next instance. We have been using the utility with this configuration for some time, monitoring each solution it presents, to confirm it is having the desired effect. We are very pleased with how it is working.

Like most of the software we have developed in-house, our stacking utility is simply an automation of repetitive tasks. Curated automation, perhaps? In all the research we have done into automation, it is apparent that layout artists can do the work much better than a machine. The solution which has proven to work most effectively is to use automation in a way that simply removes tedious steps. The stacking finder is proving to be a very useful tool for further enhancing the quality of our work.