The tantalizing problem of using "Grid System"

Lately, I have been searching for methods to effectively learn my study material. My use case is long term storage of items and retrieval in an exam like environment (contrary to a memory competition environment).

Memory Palaces (MP) are the most effortless and natural memory technique for me. They are easy to create and really easy to use, given that the images placed in loci are vivid and specific enough (also revisit those palaces regularly to dust them off!).

I have found using MP is most effective when we have the complete picture of the whole topic, at least the aspects of the topic that I need to remember. For a given topic, say “Currency Exchange Rate”, I would want to remember a lot of facts and points in a related logical order: Definition, Types of Currency Exchange Rates, Management of Exchange Rate in a country, Challenges, Reforms - Govt committees formed, Reforms - Historical perspective, and then some current factoid. It is usually not the case that I am fully aware of how many loci in a MP I will need to remember the info. I may be reading up on a few topics now and then read up on related info (like challenges) in some other topic (like Foreign trade policies). Then I would want to “expand” my Currency Exchange Rate MP to accommodate additional info. In such cases, I found Mnemonics to be really effortless and highest effort-to-benefit ratio memory technique. I can quickly create mnemonic to remember the key points and lists. All this is manageable until we have a few mnemonics. Pegged lists are better (at least for me) to remember items.

The main issue with all this is, I am finding it hard to RECALL my pegged lists. For example, I know I want to recall my peg list associated with Currency Exchange Rate - Reforms - Historical perspective. I can easily remember the “keywords” I want to use to recall. Even with spaced repetition, How can I have a “hook” sort of thing for the particular peg list that is associated with this (sub)topic?

How can I do this easily? I have been looking into “Grid System”. This sounds like a viable solution for my use. I can have a table for a particular topic and then one keyword associated with that topic. But I am finding it hard to recall the peg list. Also, the tables/grids themself become kind of blurry after having around 10 tables.

I want to study about and use this method. Unfortunately, I can’t find resources enough to fully understand this concept. Google search shows only this link:

ArtOfMemory forums have some discussions but I didn’t find any where this was described fully:

Apologies for the long post, but I tried to be as clear for my case as possible.

TL;DR Need some help in understanding and using “Grid System” for studying for an exam.


I’m not sure exactly what they are referring to, but here’s the description from that page you linked to:

In the middle ages, practitioners of the art of memory largely abandoned the use of architectural spaces – it had been a fashion of the ancients – and substituted mental grids, either square, circular, or semicircular on the model of an amphitheater. Like building interiors, a grid of rows/columns consists of regular groups of loci (cells) that can be filled with content. Unlike the various parts of a building, however, there is nothing intrinsically distinct and memorable about cells in a grid, so the key to this system is using a cross-referencing code system.

It has been many years since I read The Art of Memory by Frances Yates, but there is a chapter in there titled “The Memory Theatre of Giulio Camillo” that describes one concept of a memory palace in a artificial theater. Here’s a blog post about it:

Here’s another quote from the page you linked to:

In the grid system, the dimensions or axes of the grid – the rows and columns – are different known series, be they numerical, alphabetical, or whatever. If each cell has a row letter and a column number, that two-character combination becomes a hook for the mental image that goes in the cell. Each mental image simply incorporates the code for that cell.

… And the only limit to the number of dimensions of such a system would how many natural series of things you can use for coding. There’s the alphabet – a series 26 items long that almost everyone knows by heart – and the numbers; even just using the digits 0-9 (or 1-10) plus the alphabet gives you a grid with 260 cells – more than I’ve ever needed (but I’ve never needed to memorize a book). Multiply that by the 12 months of the year and … well, you get the idea.

Based on that, it sounds like the author uses a grid of 26 alphabet peg images for columns (or rows) and images for 0-9 (see the number shape system for an example). 26 letters times 10 letter shapes would give you a 260 location grid that you could overlay on an imaginary amphitheater or building.

So if you used the letters for columns, and they were alphabet animal pegs, like A=alligator, B=bear, C=cat…F=fox…Z=zebra, and number shapes for 0-9. then the location F4 would be “fox” with the image for the number 4 (a sailboat, flag, or whatever). You could have the fact you want to store at that location interact with the fox and the sailboat (or whatever images you use for the pegs).

[Edit: the source mentioned row letters and column numbers. My illustration has row numbers and column letters, but the basic idea is the same.]

Giordano Bruno is mentioned on the page. There is some more information about his systems here:

Someone else here might know more about it. I’m just taking a guess. :slight_smile:


Thanks a lot, Josh for the reply. I will look into the sources that you referenced in here and possibly update on how I made it work for my case. Cheers!

That looks a lot like the Tony Buzan sem cubed system. My basic understanding (This might be the way I plan to use it leaking through) of it is to imagine it like a hotel and 100 floors (which are 10x10 pegs for hundreds and thousands). Then every floor has 100 rooms (P of your PA or PAO system). This in theory gives you 10,000 pegs to hang stuff off or even one massive memory palace to put stuff into. You could easily double this by using your O as well in a separate sem cube or as 100 pegs in each room.


Woah. This sounds interesting. Have been trying out some stuff with Memory Palace currently. Thanks for pointing this out!

For anyone interested in it, there is a SEM Cubed page in the wiki.

The wiki as recommend by Josh is where I found most of my info about the system. I cannot seem to find the book by Tony that it is explained in however I have seen someone talking about it in a forum reply saying it put forward and not very well explained.

I sort of have two base ideas for memory places using this system. A hotel so you have a 100 floors and then the rooms are the P or O of your PAO. An office building with 10 floors and 10 wings on each floor with a hundred desks on a wing.

In theory that would give you 4 buildings with 10,000 locations to store 100 pegs on a permanent basis. This would be the place to sort knowledge and cross reference it I think. I think you really need six ie Geography, Entertainment, History Arts & Literature, Science & Nature, and Sports & Leisure. I could make a warehouse work I think. 10 rooms by 10 rounds with 10 by 10 shelves maybe.

I think it needs an index so I wonder if you could have something like conference rooms for TV, Poets and other things. You find Shelley in the room and he has 1792 connected to him so that the location with more information about him. You find Buffy with 1997 connected in the TV room so you go to room 1997 for more info. There is also no reason that an object or person cannot act as a sort of port key for a journey. You find an Amazon in the river room with 4000 connected and so on.

I get the Sherlock idea of London or a file room however I think one is too big and the other has no journey to it.


I have his book, Use Your Perfect Memory, which says that SEM3 is described in two other books: Master Your Memory and Memory Vision. I just searched, and there are some used copies available for sale online.

1 Like

I will definitely try to implement this idea. Thank you for it.

Thanks, found them at next to nothing this time so will give in and see what they say. From what was said before it is not very will described apart from the basic idea which seems to make sense.

1 Like

Have you had success using other grid systems? I was thinking, for example, family members and sports. You imagine each family member playing soccer, basketball, etc (obviously doing it in a different enough way to distinguish the pegs). I didn’t know if you had other ideas besides letters and numbers.

1 Like

That could work. It sounds like a person-action system. Any compound image system could probably be represented as a kind of grid: person-action (2D), person-action-object (3D), etc.

1 Like

Thanks! That’s a good segue into that topic for me.

Are these actually intended to be used as grids in the mind? That is can people actually zoom in on a precise spatial location within the 26 X 10 grid? Let’s say L6 and then P5? Why stop at 10, why not 26 by 26 etc.

I think they are really memory pegs, and the spatial location for this size of grid or larger, actually isn’t/can’t really be used.

I would buy Master Your Memory as it an actually Memory book. Memory Vision is like 2 chapters and then a 10,000 space set of lines for the SEM system.

1 Like

I think you could do that to be honest. The struggle would be having more than one list of 26 that makes sense and is easy to remember. 26 animals of every letter would be fun as 10 was a struggle.
I think there are ways in a few books of constructing alphabet lists so it would be possible. For me it would be a nightmare I think.

I think one way would be to take your alphabet images (e.g B=Batman, C=Cat, P=Pirate) and create a peg with 2 of those images. Example: Batman riding a Cat would be BC as a peg; while a Pirate killing a Zombie would result in a PZ peg) for the alphabet pegs.

Combine the alphabet pegs with your 00 to 99 major system pegs and you have a 65,000 image peg list (I assumed that no repeats would be desired (AA, BB, etc.). (Batman riding a Cat with into a tombe for example would be the BC13 file)

If you already have A-Z images, and the 00-99 major system pegs, there would be nothing extra to learn

If you went nuts and used a 3-letter code you would have 1,560,000 pegs

I think the problem would be remembering them however the professionals are using some form of 1000 PAO so it could be done. That is why I like the hotel/building idea as it gives you a journey to map to and a room to peg in.

I think my point was you don’t need to remember them.
You need to remember the major system pegs 0 to 100, and your alphabet pegs.

If you have both of these already, then you could generate about 1.5 M pegs when needed.

IF you wanted pegs based on the alphabet