# Whisky treasure-hunt: Cryptogram which refers to a place in Scotland

UPDATE 4 - Inserted original list of questions (no more, no less)

Could someone help me solve this code? I've tried for 35 years.

13 25 22 19 9 30 32 11
12 25 8 29 17

It is a place name in Kirkcudbrightshire Scotland, perhaps something like a castle or a tower or bridge or monument. This is question 4, as shown on the following sheet. The answers that are shown include National Grid reference numbers for each location.

When I was so frustrated I requested another clue of the author. Here are the three additional ciphers that I received:

19 9 25 15 22 29 30 13
14 26 10 17 24 28 25 28 16 34 27 27

27 12 26 9 20 31 14
20 14 25 29

19 25 13 30 17 24
7 17 29 15 31 12 22

If it helps, here is a website that shows about 8000 named places in Kirkcudbrightshire. These named places are listed within each of the 28 parishes.

Update: OP says correct answer for Q3 is more likely to be NX 64454 74089.

• The extended discussion formerly in the comments has been moved to chat. Please continue any discussion about solving there. – Kevin Jan 22 '15 at 4:48
• Can we get clarity on clue 6 please? Is it 749 479 or is it 749 472 ? – BmyGuest Jan 22 '15 at 20:55
• @BmyGuest - Q6 is not 749 479. Dundrennan Abbey is located at 749 475 and from the satellite view it looks like Thomas Dewar's headstone is at 749 474 or 749 475. (Most of the discussion is in chat now - link above.) – Len Jan 22 '15 at 23:07
• Elizabeth, did the author say what the three additional ciphers encrypt? If not, then there may be mileage in assuming that they encrypt other placenames in the county. Perhaps someone with a searchable long list of Kirkcudbrightshire placenames could post lists of ones that consist of two words, lengths {8,12}, {7,5} and {6,7}. There can't be many that are {8,12}. Then we can get stuck into some reverse engineering. – h34 Jan 29 '15 at 17:58
• @h34: I've looked through the list with a particular emphasis on the 12-letter word and come up with nothing that matches. That word is notable because it has two very important letter patterns : the $aba$ and the $abcc$ ending. There are several words in english that have a$abacdee$ pattern (such as "eyeball"), but none that are twelve letters long, and no placenames from the list. Whatever it is, it's almost certainly not a substitution cipher unfortunately. – Foo Barrigno Jan 29 '15 at 18:14

## 7 Answers

EDIT - Apparently the answer is "dead easy" which suggests that some local knowledge is required. So here are a few additional comments in italics and I give up.

The 6 digit coordinates are from the Ordnance Survey National Grid reference system. The first 3 digits are eastings and the last 3 digits are northings. In about 1980 when this treasure hunt was created, coordinates would have been read from a paper map with some uncertainty in the 3rd digits. For most of Kirkcudbrightshire, coordinates are entered into the "Grid Reference" box like this answer to Q1 - NX680509.

Here is some info about the other 9 questions. With a few minor exceptions noted below, all answers and coordinates appear to be correct:
Q1 - describes the inscription at the entry to Kirkcudbright Tolhouse (refer to "Rambles in Galloway")
Q2 - is explained in "Rambles in Galloway"
Q3 - describes a memorial with inscriptions near Lowran Burn - see photos on this page. The disappearing castle would be Lochivar which does not seem to be associated with the Q3 answer. Policeman could describe black and white markings similar to the band on hats worn by Scottish policemen.
Q5 - is word play with Carsfad
Q6 - describes the headstone inscription for Thomas Dewar. Coordinates are closer to 749 474
Q7 - is an anagram for the answer which is Southerness Lighthouse
Q8 - is a substitution cipher for Kirkandrews Church
Q9 - is a poem described in "Rambles in Galloway"
Q10 - is reference to minister at Kirkgunzeon - not clearly stated as 1896 but answer is most likely correct

The book, "Rambles in Galloway" may contain the answer to Q4 which could be a book cipher. Two editions were published. The first edition is available free from Google or from the British Library. The second edition is available from the British Library.

Q4 does not seem to be a substitution cipher. There are no word pairs in the list of named places that match the pattern of any of the 4 cipher pairs. The pattern of "14 26 10 17 24 28 25 28 16 34 27 27" is distinctive but no word in English matches this pattern. Even progressively removing characters from the cipher words does not help to match the named places.

Q4 does not seem to be a hidden grid reference. Addition or subtraction does not provide any meaningful grid reference numbers. Applying different rules to extract 3 digits from the first "word" and 3 digits from the second "word" does not provide a meaningful reference number.

Q4 could be a book cipher but no answer was obtained from monument inscriptions, numerous poems, or the book "Rambles in Galloway".

Using the final calculation and a grid of points across the county, the location of the whisky has to be just east of Rusko Castle, within the 3 red blobs on the following map. (The two discontinuities are caused by the grid reference system.) Coordinates for a rectangle around the lower blobs are eastings from 604-665 and northings from 577-631. The upper blob is eastings from 629-654 and northings from 671-676.

It seems that there are only 2 place names of interest within these blobs. They are a) the Watch House in Twynholm and b) Cairntosh in Girthon. Using these coordinates to back-calculate the location of the Q4 answer provides a) no points of interest near Glenarm, Urr and b) a location which is near Cally Park, Cally House, and Cally Palace in Girthon.

Assuming everything is correct to this point, the answer to Q4 should be something around Cally Park (NX607553). I would be very happy if somebody figured out the Q4 cipher.

• If you mod-26 the values (="MYVSIDFK LYHCQ" for the Q4 cipher) then do you get any place-names? – A E Jan 21 '15 at 11:06
• Does it make any sense if you look at the difference of numbers than on the numbers themselves? – BmyGuest Jan 22 '15 at 6:52
• I also wonder why the "4 ciphers" come in pairs of number sequences... ? This seems to indicate that both series need to be somehow combined... Can you multiply or add them and then interpret number pairs? (Subtraction/division won't work as the length of numbers switches between the first or the second line being longer) – BmyGuest Jan 22 '15 at 21:19
• New revision looking good, Len. How do you derive the three red blobs? – A E Jan 28 '15 at 11:55
• If the Q4 answer were "Cally Cottages", reversed, then it would be in the right location and have the right word lengths (8,5). – A E Jan 28 '15 at 12:02

I assume this is a substition cipher. I know that some newspapers pose such problems. I checked to make sure it is not a rotation cipher, which it is not.

For this cipher we have the following conditions:

1. There are two words
2. The first word has 8 letters and the second has 5
3. The second letter in both words is the same
4. All other letters in both words are different

There are several possible obvious choices for the second word: Beach, Point, Sound and Tower; and numerous other more obscure possibilities: Cairn, Carse, Cauld, Clint, Corse, Court, Cowld, Craig, Firth, Grain, Haven, Hirst, Hurst, Lough and Slack. However, none of the well-known places having these words in Kirkcudbrightshire have a matching first word that I could find. There are several common places that could be the first word: Boreland, Dalgoner, Dumfries, Duncraig, Dunragit, Dunscore, Muncraig, and Twynholm, however I could find no second word to go with these.

Having gone through all the major cities and towns of the shire and found no matching places, I am inclined to believe there is a typo of some kind in the original puzzle, a common problem with old newspaper typesetting.

• 95!!! I feel like I,m 95 when I,m shaking in my shoes knowing I at last have help but cant explain it properly or press the right letters on the keyboard. 63yrs young!! – Elizabeth Martin Jan 19 '15 at 6:32
• have sent it to you,map references may not be right for the ones solved but when I get to that stage I,ll deal with that bit. Thank You. Don't you be all clubbing together and nipping over to Bonny Scotland with spades – Elizabeth Martin Jan 20 '15 at 5:04

Credits to A E for figuring out that they were National Grid references. I tried MGRS and UTM readings, but they all went into the sea instead.

I have downloaded the NX grid map from Ordnance Survey, and plotted the 9 known locations onto the NX grid map, shown below:

There doesn't appear to be any obvious "cross" on the map, but someone with better understanding may be able to figure out something from this.

The image was generated using the following Mathematica code, where nx was the image.

point = Show[nx,
Graphics[{PointSize[0.010],
Point[{{680*4, 509*4}, {739*4, 622*4}, {644*4, 741*4}, {606*4,
856*4}, {749*4, 474*4}, {977*4, 542*4}, {601*4, 481*4}, {584*4,
604*4}, {866*4, 667*4}}]}]]

Due to stupid image size restrictions in the SE uploader, the image was heavily compressed to JPG, and was also shrunk by SE's default image host.

A full resolution PNG is available here on a temporary host. Please do suggest decent anonymous image hosts.

Rewrite on parity solution (Credit to Len for the inspiration that the boundaries can also be used to reduce the size of the solution) This rewrite aims to include additional information on Kirkcudbrightshire's boundaries, since both Clue 4 and the final whisky location are in Kirkcudbrightshire:

The fourth number of the floor of the mean of the 10 numbers is 4 (9-5).

The 4 least significant digits of the coordinates of 4 must range from 2503 to 3502 in order for the 4th digit in the floor of the mean to be 4.

Therefore, the loci of possible positions of clue 4 have been reduced by parity to 10% of the original solution space, assuming all other positions are correct.

In addition, since both the whiskey point and Answer 4 must be in Kirkcudbrightshire, the answer space is again decreased. An image mask for Kirkcudbrightshire was produced using the image here and fitted onto the main NX map grid, in order to determine whether a given NX grid reference is within Kirkcudbrightshire.

Performing the transform on the mask with this code

ksData = ImageData[ksMask];
isKs[a_, b_] := If[ksData[[b]][[a]] > 0, 0, 1];
whiskeyTransformY[a_, b_] :=
Mod[Floor[(a*1000 + b + 6451505)/10], 1000] + 1;
whiskeyTransformX[a_, b_] := Floor[(a*1000 + b + 6451505)/10000] + 1;
p4Mask = Table[
Table[isKs[x, y]*If[2503 < Mod[(1000 x + y), 10000] < 3502, 1, 0]*
isKs[whiskeyTransformX[x, y], whiskeyTransformY[x, y]], {x, 1,
1000}], {y, 1, 1000}];
Image[p4Mask]
wl = Table[Table[0, {x, 1, 1000}], {y, 1, 1000}];
Table[Table[
If[p4Mask[[y]][[x]] == 1,
wl[[whiskeyTransformY[x, y]]][[whiskeyTransformX[x, y]]] = 1], {x,
1, 1000}], {y, 1, 1000}];
Image[wl]

results in the following images for the loci of A4 and Whiskey Point (white: valid NX, black: invalid NX). Their overlays onto the map are shown:

A4 (full res)

Whiskey Point (full res)

• @AE Yes, I realised they shrank the image as it was too big (4k by 4k). Any alternative proper image hosts? – March Ho Jan 21 '15 at 11:21
• @AE I refuse to use Dropbox (requires annoying install/signup) and Picstocloud redirects me to picstocloud.com/0 once the upload completes. I have put it up temporarily on an unrelated host. – March Ho Jan 21 '15 at 11:40
• Good effort. At this point I think reverse engineering the 4th answer from the points given seems more likely than anything else. – No. 7892142 Jan 21 '15 at 12:09
• @AE Try whatimg.com/i/cm2LTz.png instead? – March Ho Jan 21 '15 at 12:38
• OP has revised their answer to Q3 so the locus of possibility will have changed. – A E Jan 22 '15 at 9:28

Some small progress, I think: the second of the three additional ciphers, namely

27 12 26 9 20 31 14
20 14 25 29

may well encrypt the words

SELKIRK
ARMS

Reasoning:

Douglas McDavid, one of the owners of the Selkirk Arms in Kirkcudbright, is known for his prizewinning puzzles and has been since he was manager of the Cally Palace Hotel in Gatehouse of Fleet, another town in Kirkcudbrightshire. (He is referred to once in the article at that link as "Duncan", presumably a typo.) The Selkirk Arms also hosts an annual "brain game", the winner of which is awarded the "egg head" prize. The said establishments, it would be safe to say, also sell whisky. I'm hazarding the guess that he is the author of the puzzle and that the second cipher encrypts the name of the establishment he currently part-owns.

Moral:

cryptanalysis is nothing without humint :-)

• This puzzle is 35 years old, and he purchased the hotel in 2007. He very well (and is likely) the author of the puzzle, but that doesn't mean that this clue encrypts the Selkirk Arms. – Foo Barrigno Jan 29 '15 at 23:29
• @FooBarrigno - Yes, but the additional ciphers may be a lot more recent than the puzzle itself. Investigation of geographical placenames hasn't yielded any candidates for {8,12}, but common sense suggests that each of the additional ciphers is Kirkcudbrightshire-themed. – h34 Jan 29 '15 at 23:35
• the additional ciphers are perhaps 5or 6 yrs younger than the original part – Elizabeth Martin Jan 31 '15 at 4:18
• @Elizabeth - Thanks for this. What can you tell use about the setter or the Kirkcudbright Squash Club at the time of the puzzle? I'm thinking that the additional clues might not stand for geographical features, but might encrypt the names of people or places having something to do with the club or otherwise important for some of its members. – h34 Jan 31 '15 at 22:24
• Kirk squash club was just starting up don't think theres anything out of the ordinary there.Mr Shamash is a different kettle of fish, keen golfing family you could google him,well heeled old Kirkcubbright family,wicked sence of humour and I bet loving the fact this is driving me mad.He,s on hols at the moment but said he,d get back to me when he gets back after I asked for an additional clue,(you saw what I got last time I asked)He said at the beginning his young son as was then got it straight away it was dead easy!!!I think it may be quotations!!! – Elizabeth Martin Feb 1 '15 at 23:57

Some bits on reverse engineering the answer:

If we can assume the 9 other clues to be solved correctly, then we can use the "sum all divide by ten" and the "check first number..." clue ( 9-5=4 ) to get at the first digit of our hint 4 solution. We should also be able to determine what the last digit of it is not if we assume that "ignore the after comma digit" hints that it is likely not zero! Ok, that's only 2 out of 6 coordinate grid digits, but it limits further searches. (That's why I would like to have the 9 found numbers be fully confirmed. The image is a bit dubious about 749 479 vs 749 472, and Elisabeth also brought in some doubt on the last "rounded" digit of Q3.)

There are 2 more points in the general instructions which make me wonder:

• if we need the sum (and maths) to get to our final coordinates,why are we required to mark the 10 solutions on the map? This implies there has to be some information when looking on the dots of the map, doesn't it?

• "Stand on one..." This seems to indicate the final destination has(/had!) something to stand on, which exists in multiple instances and is recognizable for that spot. What could that be? Are there some hills with special rocks, maybe?

Finally, "look for a tree taller than you". This might be really bad news for a 35year old puzzle! I am just glad that we all are only interested in the solution to the puzzle, and not in the whiskey :-). It might be hard to impossible to find it even with the right solution!

• I,m interested in the whiskey.I,m going to email the author again.I went to the U.k last year and had an invitation to visit him,all fell through when I broke my leg after only 4days and had to come home!!! – Elizabeth Martin Jan 22 '15 at 21:33
• @ElizabethMartin - most of the discussion is in chat now. The link is in the first comment by Kevin at the top of this page. – Len Jan 22 '15 at 23:10
• @BmyGuest - The mostly likely reason to mark the points on the map is to read off the 6 digit reference numbers (the puzzle was created in about 1980). – Len Jan 23 '15 at 23:54
• @Len Good thought.Yes, I think that is enough explanation. – BmyGuest Jan 24 '15 at 7:56

I have some serious doubts about this answer, but it's too long for a comment so here goes:

• I reverse the digits in each row of the question.

• I use the first row as Eastings and the second row as Northings.

This gives me gridref NX 11230 71928, which is a point in the hills near Drumcargo.

Reasons why this might not be the answer:

• It's quite far away from all the other clues.

• There's no major landmark there.

Reason why it could be the answer after all:

• It's publicly accessible dry land in the middle of nowhere. This would make it a good place to bury the prize. Or at least, better than (say) the middle of the Irish Sea or somebody's back garden.

• Although it's not very near the other points, it is definitely within driving distance of them.

• It's fairly close to the nearest road, which would be an advantage if you were carrying a heavy prize from the car to the burial point.

• OS map shows lots of cairns in the area so you could build a new one to conceal the prize.

Fundamental reason why this answer probably isn't right:

• Inevitably there's going to be loads of possible gridrefs that you can get by manipulating the numbers in the clue, in various ways. It's just numerology. There's few reasons to think that this particular way of manipulating the numbers to get a gridref is superior to any other way of doing it. (The few reasons are that this point is at least not in the sea and not on private property).

Reason why I've posted it even though it's probably wrong:

• If this probably-wrong answer gives someone else an idea to get the right answer, maybe that makes it worth posting.
• the hidden 5 digit number is 75823.You are all too clever for me to understand.I think if you just stick to that one clue No. 4 and not try to connect it with any of the others.I realise it must be difficult when you've no idea of the area and over active puzzling brains and I do appreciate your help. – Elizabeth Martin Jan 21 '15 at 18:25
• Thanks Elizabeth. It's true, "over active puzzling brains". :) – A E Jan 21 '15 at 19:19
• I GOT AN ANSWER WRONG AND AFTER 35YRS THE RIGHT ANSWER HAS COME TO ME.NO.3 IS A MONUMENT BUT I GOT THE WRONG ONE REF645 741 EXACT REF IS 64454 74089.wHOOP wHOOP – Elizabeth Martin Jan 22 '15 at 1:12
• @ElizabethMartin Is there some means one can use to see the monument(s) in the area? From aerial imagery it seems that the grid reference given seems to point to a nondescript piece of woodland near a river. gridreferencefinder.com/… Other than off-by-one errors, it doesn't seem to be different from the one posted in the original image. – March Ho Jan 22 '15 at 9:30
• @MarchHo, switch to OS map and you can see it - but that's very close to the same point that we had before..... ? – A E Jan 22 '15 at 9:32

Yet another (so far failed) attempt.

We've established that

• The answer the Q4 is likely some text and not a coordinate/number.
(It comes in 2 parts of different length and the additional hints following the same basic idea do not have any reason to be numbers, as they are out of connection.)

• Q4 is not a substitution cipher. (Because nothing matches the pattern)
Modifying the numbers such that they then become a substitution cipher did not yield results for simple maths, and it is unlikely that something rather involved has been used due to the other questions.

We are therefore looking for cipher which maps numbers onto letters but in a non-unique way. (Or it would be a substitution cipher.) This could be a code-table where different numbers map onto identical letters, but it is hard to envision one which would be 'obvious and dead-simple' without any additional clue.

This leaves two things jumping to mind: a flexible transposition cipher where the numbers are providing a key, or a book-cipher. Both would need a source-text to act on, and the argument of a 'dead simple' cipher seems to point rather to a book-cipher than a complicated transposition cipher.

As a result, I continue working on the assumption Q4 is some sort of book-cipher, with the numbers representing a 'picking pattern' from some sort of source. Typically, such ciphers (if meant to represent letters) either specify Paragraph / Word / Letter in a numeric way. However, we only have double-digit numbers up to 34 which seems to indicate a more direct cipher. It could be

• Pick n'th word of source (and take first letter of it). Source at least 34 words long.
• Pick n'th letter of source. Source at least 34 characters long.

The second option seems more straight forward to me, but it would also suggest that the 'source' would be something the puzzle-solver would rather easily guess as there are no other hints to a source.

I've therefore written a little script which takes the code-indices and a source string and outputs the according 'pick n-th letter' solution. It does this in 2 variants: with and without counting spaces as letters.

The task now is to try out as many 'potential' source-strings as possible. I've so far only used

• Kirkcudbrightshire squash club donate 1\$
• Somewhere in Kirkcudbrightshire is a buried case of whiskey

Both did not yield something useful. It would be great, if we could together compile a list of "source strings" we might want to investigate. In particular Elisabeth, as she might think of "obvious" source-strings due to local or historical or personal knowledge. Are their some mottos or inscriptions etc, which would be useful? They can't be too short, tough.

• @ElizabethMartin Could you think of a couple of 'source' texts which could be useful to be analyized? – BmyGuest Feb 5 '15 at 12:42
• Considering the fact that all the ciphers have two lines, the key for the first line could be "Somewhere in Kirkcudbrightshire..." and the key for the second line "Keeping a record..." – Aravind Feb 5 '15 at 14:39
• @Aravind You're right about that, but because I check the full lines with any key, I would get some "sensible" answer for the first part anyway - which I don't. – BmyGuest Feb 5 '15 at 16:29
• Hmm. What about anagramming afterwards? – Aravind Feb 5 '15 at 16:33
• Wikipedia has this list: en.wikipedia.org/wiki/… which may be useful for anyone searching for buildings with 12-letter names. – Aravind Feb 5 '15 at 16:39