How to know a linked chain is over?



When using the linking system, have you ever faced the problem of not knowing if the chain is over or not? If so, how have you solved this problem?

What I mean is: suppose you’ve linked 6 items A->B->C->D->E->F but you don’t necessarily know beforehand (when attempting to recall it) that the whole chain is composed of exactly 6 items; so when you go one by one checking each image, there is always the possibility that you think the chain is over but that’s only your memory that is faltering – or, you might think the chain goes on but, in reality, it is over.

Using a “marker-image” at all your last images will clearly yield an unacceptable number of repetitions. You could instead repeat the first item at the end of the chain so that the completed cycle would signal that it is over, but in my case, it wouldn’t work. I have “branching chains” so the truth is that I also want to know if each image amid the chain originates a new branch or not. This means that to use the aforementioned approach I would need to link ALL non-branching images to themselves (!) which is obviously not ideal.

Thoughts? Experiences?

I thank you all in advance.

And please forgive me if I haven’t expressed myself clearly enough – that’s a weird problem to explain.




You should know you reach the end of a chain because the story you are creating for the images should end. If there is no more to the story, it has to be the end or you are just forgetting some information. The main issue that can happen is just forgetting something in the chain, but this is fixed with practice.



I agree partially with what Parkouristx said. Usually, I link each image doing something to the next one and the fact that an image doesn’t seem to be doing anything is enough for me to realize that the list is over.

I’m not sure what “branching chains” would entail, but I think I have an (admittedly inelegant) solution to your problem. Instead of a “marker image” to signal that a branch has begun, you could have an image that tells the length of a branch. Clearly, these 10 images(or 110) would have to be unused in the chain itself and so should they be something obscure. You could start every branch with one of these markers and if you see another in the link, you know the link “branches”.

The advantages being it solves the problem of repetitive markers and tells you the length of the chain, so you know if you’ve forgotten something. However, I feel that it would quickly prove hard to accurately gauge the length of especially long chains and you would have to modify the marker everytime you add or remove an image. Also, incorrectly recalling(or forgetting) the marker would prove disastrous.


Another approach might be to always use a particular kind of scenario to connect the last item.

For example, you might always end with something destructive being done to the last item by the second last item. (Doesn’t have to be destructive, could be a happy ending). Let’s say you are trying to memorize the following chain of words:

Ape, Boat, Castle, Dancer, Elf, Fox

You might memorize by imagining an Ape driving a speed Boat up to a Castle where he sees a Dancer trapped in a tower. He wants to rescue her, but there is no way up, so he enlists the help of a magic Elf to make a magic staircase, but then as he tries to ascend he encounters a giant Fox guarding the staircase. So he kills the Fox most violently.

(Note: No animals were harmed during the production of this mnemonic story. All fight scenes were done using professional stunt doubles or special effects.)

The point is that the violent end of the Fox might signal the end of the chain. This would mean you would always have to avoid using violent images except for the last item in the chain.

The deeper point is that you could use any theme to signal the end. Maybe you prefer all your stories to have happy endings, so maybe marriages and love scenes would be better endings for you.

For some people, their linked chains do not make an actual story, just a series of links. In that case, you could reserve a specific set of action verbs that always signal the end of a chain.

Hope this is of some use.




Thanks to you both for your input (Parkouristx and nkp).

The problem with Parkouristx’s argument and, to some extent, to nkp’s first comment is that they are both rather cyclic. When you forget a link (i.e. the next linked image), the previous one (i.e. the last you remembered) will generally seem to be doing nothing; at least nothing beyond what it has already done to/with the image before. So, the chain will seem to be at its end. It is a cyclic argument because you are telling me that if there are no more images then I’ve reached the end; but no more images IS the definition of the end — the problem is distinguishing between the actual end and forgetfulness.

The “practice argument” is not enough either. Anything with enough practice gets “fixed”. That’s because it becomes natural memory or an instantaneous reflex of the body if we are talking about mechanical skills like typing. But the whole deal with Mnemonics is to quickly and effectively acquire a massive quantity of information in an interval of time insufficient for natural memory to do its magic.

Moreover, there are “stories” and there is simple linking. As many have written about it and I quite recently did on this thread, these are similar but different animals. With pure linking, it is even harder to figure out the end because there is no causality across links. And a story is not a romance story where the good guy gets the pretty lady at the end; this is a non-sensical story with no rime or reason that doesn’t necessarily (almost never) end in a logical way or with red curtains coming down and a big “The End” on the screen. The story just reaches the last image and that’s it.

Now, the “length image” idea is a nice one (Thanks!), and I admit I have been tinkering with it for quite a while. Yes, the need to modify the marker when changing the size of a chain might be a problem, but for more “stable” projects it would not be too big (now that I am rereading this, markers would be even more important for “unstable” projects; but let’s leave this problem for later…). And obscure images wouldn’t be a problem either because, generally, my chains do not include numbers, so my 00-99 images could be used as markers with not much confusion. The real problem is two-fold: the repetition of markers and the need to count the images. By using length markers the repetition is greatly reduced, but even so, when you are talking about hundreds or even thousands of chains, then it quickly gets boring, thus prone to forgetting. And, although I love to count cards in BJ, I really don’t feel like doing so when recalling general knowledge. And doing it backwards when reaching the length marker simply sounds counterproductive.

I am probably pushing it here, I know. I’ll keep thinking about it and, hopefully, someone might appear with the perfect solution, who knows.

Thanks again.



Hi, tarnation

Thanks for your input. I am glad you pointed out in the end that not all chains are stories, but I get your point either way.

Yes, that’s a good idea, but I am still worried about repetition. Like you’ve done to your “letter-based system” a while back (If I recall correctly), I could have a list of possible markers or contexts or actions that would trigger the end of the chain, but for some reason, I simply don’t like these solutions too much. I feel chained by them (no pun intended). And that’s coming from someone who wanted to define strict image placement to represent sentence structure, huh?

But, like I said, I am probably pushing it here. It is very unlikely that any solution will be able to dispense with some sort of image marker and hence with the ever-present problem of repetition.



P.S.: Stunt doubles or special effects are not as memorable as the real thing :wink:


Mnemoriam, if you could give an example of what you’re trying to memorize it might help?

Anyways, now that I’ve thought about it a bit more, I’ve come up with a better idea. I agree that it’s most likely that you will have to use a type of “marker image” in some form. So, why not have the marker be itself linked to the chain it points to?

My reasoning goes like this: If your chain is branching, it’s doing so for a reason. Let’s say you’re memorizing a list of genus(genuses? I mean the biology classification) and the species under them. You have a genus branching into two lists of species: extinct and non-extinct. You could have a beating heart signal the start of the non-extinct branch and a grave begin the extinct one. Maybe if you wanted to add an endangered category you could have a man surrounded by swords pointed at him.

Now, I realize the example doesn’t exactly point an optimal way of remembering the list of genus and their species, but I was unable to come up with an example where branching was the better solution.


Hi, nkp

Thanks a lot for putting such an effort into this problem. Your example is perfectly fine, so I’ll use it to illustrate the problem. But, before that, just so you know, it is not as much what I am trying to memorize as it is how I am trying to memorize. I am working on revamping a method I have been using for a while to memorize pretty much anything. To make a long story short, it’s just a way to create and memorize hierarchies based on any input.

With respect to your idea, I think that you are now answering the other problem I quite unintentionally put forward in the initial post. The thread is about how to signal the end of a chain, but I ended up asking how to signal the beginning of a “branch-chain” as well, which is, if I understood you correctly, what you are answering now.

So, please, look at the image below while reading the rest of the post:


The black arrows show how we would usually visualize this hierarchy. The blue arrows, in turn, show how I link them mnemonically. If you pay attention, no individual image has more than 3 links (1 incoming; 2 outgoing). On each level (like Genus or Species), you have a normal linked chain where each image doesn’t have more than 2 links. But in order to “branch into” a new level, I need a “triple link” (highlighted in red for your convenience).

What you are saying is that the image “Genus-3” should have both a heart and a grave attached to it in order to signal me this image has two branches while at the same time telling me their themes. Is that it? If I interpret you correctly, this is the “ordinary” way of imagining hierarchies; if Genus-3 would branch into 10 categories, I would have to add 10 markers to the same image. But that wouldn’t be ideal, at all. But, by considering hierarchies the way I do, that’s actually a nice idea. I would only need to add a grave no Genus-3; that way, I would know it branches, and I would enjoy the added bonus of knowing the theme of the branch.

However (and that’s a big “however”), that is exactly what I would be doing anyway. In order to branch into any new level, I have to conjure images corresponding to each item of this level, being these (the items) my final to-be-memorized images (say, the names of the species), or being them just a category label to unite the items below (like “extinct” and “non-extinct”). So, my images, in this case, would be a grave and a pumping heart anyway (given that I had a nice imagination such as yours), which I would have to link to Genus-3. The only difference in my case is that I would link Genus-3 to the grave (while also linking Genus-3 to Genus-4) and only then link the grave to the pumping heart (while also linking the grave to whatever the image of Species-1 is).

Do you follow me or it is I who didn’t follow you at all?

In any case, I thank you very much for helping out. I still haven’t had much time today to think through yours or tarnation’s ideas, but I’ll do so soon and I’ll be sure to post here any follow-up ideas I might have. Do so too if you will, but, please, don’t spend too much time thinking about my crazy ideas!



Yes, you understood what I meant perfectly. I’m afraid I’m out of ideas, but I’ll let you know if I do come up with something. In the meantime, please let me know if you figure out something.


Just mangle the last item in the list.
However, if you’re looking for a system that would totally eliminate natural memory, I believe that would be impossible. Without natural memory there would be no memory whatsoever, including mnemonics.


Hi, Celtic

Thanks for this. It seems that this kind of solution fits most people, but, unfortunately, not me. I find mnemonics a difficult thing; adding markers to images is hard for me. My images already have too much attached to them, being it related to the meaning per se or to the link structure of what is being memorized. Adding extra information that repeats itself many times and is unrelated to the subject-matter (actions or objects relating to “mangling”, for example) is something I try hard to avoid.

Searching for a method of memorization that doesn’t use natural memory would be like trying to write without using ink (or bits, for that matter). Of course, that’s not what I intend. However, I believe the whole idea of mnemonics is to ease the burden on natural memory as much as it is reasonable. In the way I do things, I spend most of my time first summarising and structuring reading material, then conjuring images and linking them. But these are primarily to enable the memorization of the structure of the material and only the very gist of the final pieces of information I want to store. Most of the information per se I leave for my natural memory. That’s why I try my best to optimize what I will actually memorize mnemonically.

That said, this whole thread has helped me find the solution I felt existed but had been blind to – it had been laughing at my face all this time.

I don’t think people here have been using mental hierarchies to remember like I have – and my solution only applies to such a method – but I will briefly describe it here to at least give some closure to this thread. If it fosters further discussion, it will be, of course, very welcome.

The simple (and, if I may say, quite elegant) solution is to link the last image (node) of a chain to the next unvisited image in a depth-first-search traverse of the tree. It might sound complicated, but it’s quite simple, actually. The “soul” of the solution came from nkp’s comment: “If your chain is branching, it’s doing so for a reason”. An example might help:

The image on the left shows the links I would create to memorize such a tree. So, the node B would link to its right sibling C (both on the same “Green” level) as well as to its first child F, in the Orange level below. F->G->H is then a normal chain. My problem was how to be sure H is the last node of this chain. Well, like nkp noted, trees are like so for a reason. I memorize just like I read: in layers. What I do is to first get used to each level in turn. When I then begin memorizing the Orange level, I will have already acquainted myself with the Green one. So, if I add the C image to the end of the FGH sequence (as shown in the image on the right), I will be accomplishing two things at once: (i) I’ll be marking the end of the chain, easily seen either by the change in “theme” of the images or simply because I am used to that image already; and (ii) I’ll be reinforcing the images of the Green level and exactly locating that Orange branch within the super-structure.

Therefore, although I am in effect increasing the length of the chain to be memorized, I am not wasting my time or effort in memorizing unrelated information. On the contrary, I am reinforcing the recall of the “knowledge-tree” which structure I want to remember, and which has now become a closed circuit. As you can see by the other red arrows in the image on the right, every time I reach the end of a chain I link it to the next image one level up until I am at the root of the tree again. The complete path you would follow if you traversed the tree first moving down the levels and then following the red arrows is one of the commonest ways of traversing trees in Computer Science, the “depth-first-search” approach, hence, my comment two paragraphs above. And, although when first learning a tree I use the other well-known traverse method, the “breadth-first-search” approach (meaning I first memorize each level entirely before going down to the next one), what we want in the end is to get to the “leaf nodes”, where the real meat of the information is stored. So, the “depth-first-search” is indeed what we want to achieve (until, of course, our natural memory allows us to go directly to the leaves without the need of any intervening levels).

I hope I’ve made myself understood, but either way, I am glad to continue the discussion if there is any interest.

Thank you all for this very useful thread (at least, for me), and special thanks to nkp.



I don’t think mangling something adds extra symbols or burden to an image. It’s just as easy to imagine a mangled car for example as it is to imagine one in an undamaged state.

Also there is no highly repetitive image or marker since it would be the actual last image of the linked list that is mangled.

I think mangling the last item is more efficient than linking back to the layer above and less likely to be confused. Knowing that the list ends because it is linked to the previous level is actually using natural memory. In which case, just use natural memory to remember list ends, or mangle the last item