While thinking of the idea of Chimera pieces, and particularly of Derek Nalls' Zig-Zag pieces, I came up with one that could appropriately be called a Dragon. Since that name has already been popular, other possibilities are Winged Serpent or Chinese Dragon, for the idea behind the piece is the long serpentine Dragon depicted in China, not the western type that loves to hoard treasure. Wyvern is another possible name. There are four static variations on this piece, and a rotatable version that could switch between any of these four. There are also eight static subvariations of this piece and a rotatable version that could switch between the eight. The basic idea is to have a linerider along a single orthogonal or diagonal axis, which can move to either side as a Knight. The available Knight moves would be those that do not end up adjacent to the piece's axis. For an orthogonal axis, these would be ones that can be reached with an orthogonal move perpendicular to the axis, followed by an outward diagonal move. For a diagonal axis, these would be ones that can be reached with a diagonal move perpendicular to the axis, followed by an outward orthogonal move. So, whatever the orientation of the piece, it would have up to four possible Knight moves. The axis along which a Rook or Bishop move is possible would represent the serpentine body of the Dragon, and the Knight moves would represent its wings. For simple record keeping, I could call the four static versions the Dash Dragon, the Pipe Dragon, the Slash Dragon, and the Backslash Dragon. These names are based on punctuation marks that correspond with the orientation of the piece, but they may not be poetic enough for an actual game. I'm thinking of two different versions of the rotatable version. In one version, it may rotate to any other orientation after moving, or in place of moving. In a slightly weaker version, it could rotate 45 degrees either way after moving, or rotate to any other orientation in place of moving. Rotation should not be allowed before moving, for that would make the piece equivalent to an Amazon. If we conceive of the Dragon as having a head at one end and a tail at the other, we could have eight static types with only one radial direction of movement. A rotatable version would be able to switch between orientations after moving, or in place of a move. Since allowing this piece to rotate before moving would make it an Amazon, this would not be permitted. Stronger variations, in which the piece has Nightrider moves instead of Knight moves are also possible.

I think "subset" would seem the most natural to me.

After sleeping on it, I woke up feeling that subset would work.

I would also personally place the Silver General and company to be in this group rather than Chimera.

With respect to the pieces in both Shogi and Chess, these are both subsets of the King. However, when I adapted these pieces for Hex Shogi, I conceived of the Gold General as moving like a Wazir or forward as a Ferz and of the Silver General as moving like a Ferz or forward as a Wazir. These are accurate descriptions, which may both be regarded as Chimeras of the Wazir and Ferz. But it is also true that the Hex Shogi versions are still subsets of the Hex Shogi King.

If we're to put these into a Subset category, then this would work best if the Man were put into the Simple Leaper category rather than the Leaper Compound category. Is that something you would be comfortable with?

If we're to put these into a Subset category, then this would work best if the Man were put into the Simple Leaper category rather than the Leaper Compound category. Is that something you would be comfortable with?

Yes, that's where I would put it.

The Man/King is not a simple leaper. That term is reserved for pieces where all moves are rotated or mirrorred versions of each other. So the King is definitely a compound.

Compounds can have subsets to, though. Which then can be compounds of subsets of simple leapers. So I don't think that we can require a piece must be one or the other. Subset merely is an extra qualification that can be applied to simple leapers and compound leapers alike. (Or to sliders, for that matter.)

So there doesn't seem anything wrong with saying Gold is a subset of the King. That is more true than just saying it is a compound, as it is really a compound of subsets. That the King is a compound should be considered common knowledge. King and Queen have special status in this respect, because they are orthocox Chess pieces, and people can be expected to know them. Betza notation expresses their moves as a single capital, K or Q, officially 'shorthand' for WF and RB.

The large Shogi variants have many pieces, and they are virtually all subsets of the Queen. (As oblique moves are virtually non-existent in Shogi.) Some are even subsets of the Rook or the King.

What I dislike about the term 'subset' is that it has a well-defined meaning, which would consider the R4 a subset of the Rook. It seems that we now want to deviate from this meaning, by not allowing it to describe range restrictions, but only complete absence of a move in a certain direction. I think that such a distortion of the meaning of well-known terms is inadvisable.

I guess I am persuaded that a Man is a compound, since a Queen is, but that a Gold General can still be considered a Man/King subset.

I'm not persuaded, however, that an R4 would be a R subset. It depends what we are referring to as the items in the set. I consider it a set of move capabilities, not a set of potential destination squares. And I consider a range-limited slide a different move capability than an unlimited slide. I certainly don't consider a dabbabahrider a subset of a rook.

Yes, I agree. I've updated the modified date so it will appear in this weeks updates. I usually do that when I'm unhideing a page if it hasn't been edited in a while. The page owner can also force it by making some edit.

I'm stupid. I registered when I saw your post yesterday, and today I forgot to connect. Friday 6PM, end of a hard working week, I started to relax and I forgot :=( They sent two mails to ring up, but they went to my spam box. Aaaargh.

Only satisfaction, I had sent the information to some friends, and one friend was able to listen. He told me that I have not missed much.

I was shoveling snow this morning. Got something to eat an hour before start, and fell asleep on the couch from about 12:30 until a little after 14:00, neatly missing it myself. Maybe the 3 hours of sleep last night can take the blame. :)

Thanks for posting. Maybe it'll show up on youtube or some obscure website. But it was about western chess only. Looking at the Silk routes and the various forms of chess found along them gives bits and pieces of a few stories. The eastern chesses are fascinating in their similarities and differences. And the biggest mystery is the disjunction between western chesses and eastern chesses. The eastern chesses are obviously a family, and western chess is just as obviously a closely related but different family. I just want to know when, where, and how the original idea split into 2 related families. And how, of course, Japanese chess arose.

Joe, http://aworldofchess.com

Good grief! I walked right into that, didn't I! Thanks for the information. The short video was informative.

How many games are you playing to get these percentages?

"I realized that the knight is stronger than a bishop." Do you mean that? I had the feeling that on a 10x10 the Bishop is the strongest.

He means the knight in Apothecary Chess, which is augmented, is stronger then the bishop. There are two different Apothecary games. In one, the Knight is augmented by a non-capturing (3,2) zebra move. In the other, it is augmented by non-capturing (2,2) and (3,0) moves.

@Jean-louis Cazaux In apothecary chess Clasic the knight may also move as a zebra but not capture. In apothecary chess Modern the knight may also leap 3 orthogonally or move as an alfil, but not capture. These are too strong enhancements. I am trying to get to something that will keep the knight and bishop equal.

@Greg Strong For the alfil enhancement I used 512 games. For the other two I've stopped around 100. That is beacuse I had though the results to be weird.

I think it is likely that the results are weird because at only 100 games the error margin is high. Testing is time-consuming, so it is tempting to stop a test early based on initial results, but if you do that you do not have an accurate measurement and are responding to noise.

Sorry, I missed that important point. Btw, no "e" in my name :=)

@Greg

I don't think 75% can be explained by random noise, for something that was meant to be around 50%.

He dropped off without explanation after updating the graphics on his pages to use the diagram designer and shortcodes. He has not been in contact with me, and he has had no known web presence apart from his activity on this site.

Seems that creating directories for new variants doesn't work right now.

Error message:

Failed to create the directory /home/chessvariants/public_html/play/pbmsettings/cascade_chess.

Seems that creating directories for new variants doesn't work right now.

It should work now that I have put apache and chessvariants into each other's groups.

I have definitely heard about this kind of Chess variant, and I remember to have played it in Germany when I was young and not fully introduced to FIDE rules. It is a kind of popular chess variant always flying below the radar, a bit like "Queens Left Chess" with a point-symmetric setup of the pieces. Interestingly, it is the first question in the Rules of Chess: Pawns FAQ on this site.

When I restored the site, MariaDB 10 kept losing connections. To get the database to work, I replaced MariaDB 10 with MariaDB 5. In doing this, I manually edited the database backup to make it compatible with MariaDB 5, and one change I made was to remove all the FULLTEXT keys, which were what enabled it to search through the text of comments. I expect the code that exists for searching the text of comments uses special features of MariaDB 10 that are not a part of MariaDB 5, and that is why your search did not work.

Okay, I've updated the code. Try it again.

I have added the round 3 matches!

Ok, so the round 3 has begun!

By the way, what is the result of the second round?

Round 4 games have been added. Good luck!

Renniassance Chess has pieces that can do either

Thank you. I think this was the one I was thinking of. I didn't implement RennChess because of this ...

As for pieces that can only do slide‐then‐step and not the reverse, the only one that comes to mind in a game is the Transcendental Prelate.

I don't know of anything else using this piece. Team-Mate Chess uses a "Mortar" which has AG moves, but not DH. That is the closest I know of.

Despite the nice geometric move pattern this piece seems to be unemployed before Musketeer Chess. I can understand why: It creates triple forward forks into the rank behind the pawn line and is a very dangerous attacking piece. Creating a playable game with this piece is definitely a challenge (I haven't examined Musketeer Chess in this respect).

This is true.

Well, although I'm not sure it is so bad if the board is large enough.

Has everybody seen my post about round 5 beginning?

I have

Round 4 already finished? You're getting fast guys :) (joking)

I'm back for the round 5.

So far the classification:

erik 7 points

Daniel 6 points

Aurelian 3 points

Oisin 0 points

Indeed!

Hello, I would just like to apologise for disappearing and letting all my games flag, but I should be back to playing properly now.

Per Joe's excellent suggestion, this thread is for discussion of other places on the web where chess variant enthusiasts explore, play, and discuss variants.

Has anyone made link pages for these sites?

I have a preset for one of my new games. Once after saving the preset I could not access it anymore as I get a totally white window. I use microsoft edge. What happens?

My best guess is that your code contains an infinite loop. You can figure things out more precisely by going here:

https://www.chessvariants.com/play/pbm/play.php?game=Grand+Apothecary+Chess+3&settings=Default&submit=Edit

1. Save each block of code to a text file on your computer.
2. Delete each full block of code in turn until you determine which block of code is causing this to happen.
3. Restore your code from your backup.
4. Working with the block of code that is buggy, delete or comment out large portions of it to determine where the problem lies. Make sure that each portion you delete or comment out can be safely removed without breaking up control structures.
5. Once you isolate the problem to a large section, delete or comment out smaller parts of that section until you find the line that is causing your problem.
6. Examine that line to figure out what the problem is, fix it, and see if it fixes your problem.

@Fergus It seems that the infinite loop was caused by the following code:

setelem op .JW + elem .JW op 1; setelem op .jw + elem .jw op 1;

.jw is the wazir camel piece and I had wanted to to increase the number of them in reserve. Probably the dot is causing trouble. Do you agree?

@Fergus It seems that the infinite loop was caused by the following code:

setelem op .JW + elem .JW op 1; setelem op .jw + elem .jw op 1;

That's two lines of code. In the first line, you are trying to increment #op[".JW"] by one, and in the second, you are trying to increment #op[".jw"] by one.

.jw is the wazir camel piece and I had wanted to to increase the number of them in reserve. Probably the dot is causing trouble. Do you agree?

Yes, it looks like it is causing trouble. In a test I ran, it incremented #op["JW"] and #op["jw"] by one instead of the elements you named. This is probably because the period is now used as a separator between array and element names, and the setelem command now just concatenates the array and element names by putting a period between them, and uses this concatenated name with the function for setting a user variable.

What I would recommend is to follow the instructions I have previously given for replacing piece labels without using aliases, which is in the comments for the Game Courier Developer's Guide. This would give you nice clean piece labels that you can use in your code without running into problems like this.

The comments section of the developer guide does not work on edge!

Works now. Don't worry.

I get this error: Syntax Error on line 150

The last expression is 1. In for or foreach, it must evaluate to an array.

You have incorrectly defined your Mortar-Range function as follows:

def Mortar-Range leaps #0 1 1 or rays #0 1 0;

The or operator cannot be used to merge two arrays. It returns a Boolean value. That's why it gave a value of 1 where an array was expected. Use the merge operator to merge two arrays:

def Mortar-Range merge leaps #0 1 1 rays #0 1 0;

Thanks!

Fergus, is there a way to customize the blue color that highlights possible moves?

Fergus, Is it not that in the fairy chess include file the functions :

def White_Pawn-Range array where #0 0 2 where #0 0 1 where #0 -1 1 where #0 1 1; def Black_Pawn-Range array where #0 0 -2 where #0 0 -1 where #0 -1 -1 where #0 1 -1;

for generality they should be:

def White_Pawn-Range array where #0 0 var fps where #0 0 1 where #0 -1 1 where #0 1 1; def Black_Pawn-Range array where #0 0 - var fps where #0 0 -1 where #0 -1 -1 where #0 1 -1;

?

You were correct in identifying the problem. The range functions for the Pawns did not take into account the value of fps. However, replacing 2 with fps would not allow for first Pawn moves that are shorter than the maximum possible distance. For example, if fps were set to 3, this would include 3-space moves, but it would not include 2-space moves. This is what I ended up with:

def White_Pawn-Range mergeall where #ori -1 1 where #ori 1 1 filter lambda (onboard #1) aggregate lambda (where #ori 0 #0) range 1 var fps =ori;
def Black_Pawn-Range mergeall where #ori -1 -1 where #ori 1 -1 filter lambda (onboard #1) aggregate lambda (where #ori 0 neg #0) range 1 var fps =ori;

These use aggregate to create an array of spaces a Pawn may advance forward to, and they use filter to weed out non-spaces that are out-of-range. They use a named parameter in order to use its value in the lambda function used by aggregate.

I have since modified the functions above to use the filter on all results:

def White_Pawn-Range filter lambda (onboard #1) mergeall where #ori -1 1 where #ori 1 1 aggregate lambda (where #ori 0 #0) range 1 var fps =ori;
def Black_Pawn-Range filter lambda (onboard #1) mergeall where #ori -1 -1 where #ori 1 -1 aggregate lambda (where #ori 0 neg #0) range 1 var fps =ori;

Finally, I replaced aggregate with an expanded version of the values operator. With a lambda function and an array, it will work just like filter except that it returns a sequential array of each value calculated by the lambda function. Unlike aggregate, it will include values of zero, because it does no filtering on the results it calculates. As it did before, it will return the values of an array if that is what's passed to it. If it does not get a lambda function or an array, it will return the rest of the arguments following it as a single array.

def White_Pawn-Range filter lambda (onboard #1) mergeall where #ori -1 1 where #ori 1 1 every lambda (where #ori 0 #1) range 1 var fps =ori;
def Black_Pawn-Range filter lambda (onboard #1) mergeall where #ori -1 -1 where #ori 1 -1 every lambda (where #ori 0 neg #1) range 1 var fps =ori;

Fergus, What about the ability of changing the blue color of the highlighting when showing possible moves.

@Fergus, I think the berolina pawns have the same problem that the regular pawns had!

@Fergus, A third thing if I may ask in these game the king may castle in 2 ways with the rook that starts near the corner of the board or with the corner cannon. Omega chess implements a castle move with a non corner piece. Can I combine this with the regular castling subroutine in order to easily make my castle subroutine work?

I've been working on redesigning the headings so that one can intuitively tell one level from another. Besides decreasing in size, they also differ in style.

Heading One: Centered

Heading Two: Double Underlined

I had originally just used underlining, but I changed it to double underlining to avoid confusion with links, which are normally underlined with a single line.

Heading Three: Normal

Heading Four: Italics

Heading Five: Small-Caps with Overline and Underline

Because small-caps are more compact, they are used for lower headings rather than for higher ones. The use of lines parallels the use of lines in the second heading. Because small-caps all have the same height, words have a more straight appearance, which works better with the overline.

Heading Six: Small-Caps

Like the change from heading two to heading three, heading six omits the lines used in the heading just above it.

Here is why I am using small-caps for headings five and six instead of for heading two, as I used to do in Game Courier:

Using small-caps reduces the width of the line in Courier Prime even though it is not proportional.

Using small-caps reduces the width of the line in Courier Prime even though it is not proportional.

I adjusted the margins of the headings to be the same, and I put more space above a heading than below it so that each heading is closer to the text it is a heading for. But I also avoided making it look crowded by putting a heading too close to the text following it.

As far as I am concerned this header business is a large step in the direction of uglifying the website. The typeface makes the articles look like typewritten manuscripts rather than published articles, and having underlining in the headers looks weird and unproffessional. On a scale of 1-10 I would judge the old header styles perhaps as 8, and the new situation as a 3. And this new (fixed-spaced) typeface now also appears on HTML buttons...

I have to say, I think these huge, bold, double-underlined headings look absolutely horrible. Please change it back. Or, at a minimum, get rid of the underlines.

these huge, bold, double-underlined headings look absolutely horrible.

Wavy underlines, which I already tried out, look much worse.

Please change it back.

It's important to distinguish each heading level by a different visual style. I am now trying out a style used on Wikipedia for H2. It puts a full-width line underneath the heading. I have done this with the bottom border. So, the line no longer gets broken up by characters with descenders.

Since pages were looking too cluttered with horizontal rules at the top, I took some steps to reduce this clutter. First, I removed the H2 heading for the introduction section in member-submitted pages. Second, I changed the notice that sometimes appears at the top of a page to a paragraph with the notice class instead of text with a horizontal rule above and below it.

To prevent the H1-H6 styles from being used in the wrong place, I limited these styles to headings with an article, or a section within an article, as the immediate parent. In other contexts, such as comments, footers, nav bars, or tables, these styles will not be used. They may be used in comments by enclosing them in an appropriate parent element, like so:

The typeface makes the articles look like typewritten manuscripts rather than published articles,

Courier is being used only for headings, not for the body text. I find that it looks better than Noto Sans, and it works better with horizontal lines. When I tried Noto Sans, its descenders touched the horizontal line beneath the H2 heading, and the lines in the H5 heading were not the same distance from the text. I also dislike how barren and uninteresting sans serif fonts look when used for headings. I find that using a non-proportional font for the headings makes them stand out a little bit better and distinguishes them from the body text more.

and having underlining in the headers looks weird and unproffessional.

This is a hobbyist site, weird is a personal opinion, and there are only so many different ways of distinguishing six levels of headings by style.

I temporarily tried Lexend for headings. It looked better than Noto Sans, because it had a geometric design like Futura rather than a grotesque design like Helvetica. However, it did not have an italic style, and using weight to distinguish the heading styles didn't distinguish them enough.

I think I'm fine with the (pseudo)underline for h2, but the under/overlining of h5 makes it stand out more than h3-4.

By the time someone gets to h6, I don't think much distinction really needs to be made; boldface but otherwise p-style would be nearly enough. (Since you pointed out wikipedia's, note that h4-h6 are all the same, and h3 only differs from those in font-size.)

I think I'm fine with the (pseudo)underline for h2, but the under/overlining of h5 makes it stand out more than h3-4.

It works for me, and the only page I know of with H5 and H6 headings is the Game Courier Developer's Guide, which I wrote anyway. If I'm to do away with the lines, I need an alternative that works as well, and I don't have one. I did try small-caps for H5 and small-caps plus italics for H6, but italicized small-caps differs from small-caps merely by being slanted. So, I chose not to do it that way.

Since you pointed out wikipedia's, note that h4-h6 are all the same, and h3 only differs from those in font-size.

Yes, I'm not a fan of how Wikipedia is styled. I borrowed one idea from Wikipedia, because it seems to work, but I don't like how some of their headings differ only by size. It makes it harder to tell what level something is.

I am trying out a set of headings styles that don't use any horizontal lines. Instead of making all the styles completely different, I am using sharp differences in size between headings using the same style. H5 has the same style as H3, and H6 the same as H4, but each is 0.55em smaller.

After looking at some pages, I decided that the pseudo horizontal rule works better for H2. It more clearly sends the message that this is a major section. Since Wikipedia does it, it should be familiar to people, and I don't think it would be considered an unprofessional look.

I switched back to using Noto Sans for headings on the main site, though I will use Courier Prime for headings in Game Courier for thematic reasons. I could not find any good substitute for Noto Sans among sans serif fonts. I wanted one that put bars on the capital I, and there were few of those. I tried out IBM Plex Sans, but it didn't look as good. I figured out how to keep letters in Noto Sans from touching the horizontal rule by using some bottom padding. Since I'm no longer using small-caps or underlining and overlining, H5 and H6 no longer look as bad as they used to when I was trying out Noto Sans. With these problems taken care of, I figured it was okay to go back to Noto Sans.

It's important to distinguish each heading level by a different visual style.

This is a hobbyist site, weird is a personal opinion, and there are only so many different ways of distinguishing six levels of headings by style.

Wavy underlines, which I already tried out, look much worse.

Fergus, I'm pretty surprised at what I'm reading.  You have suddenly decided, unilaterally, that six levels MUST be displayed so differently that it slaps you in the face.  And, if the double underline looks terrible, hey, at least they are not wavy?  Is that really your answer?

I think you would agree that I have complained about very, very little all these years, despite several large issues worthy of complaint.  I'd rather not go into enumeration, but I will if you doubt it.

The fact is 99% of the content on this site was not created by you.  You cannot just do whatever you want with it.  The huge, bold, double-underline type is categorically unacceptable to me.

You might want to read the later comments before replying to earlier comments that are no longer pertinent. If you still see the double underlines, you need to clear your cache.

After looking at some pages, I decided that the pseudo horizontal rule works better for H2. It more clearly sends the message that this is a major section. Since Wikipedia does it, it should be familiar to people, and I don't think it would be considered an unprofessional look.

OK, this is better. Not necessarily better than last week's styles, but at least not far worse. The typeface no longer sticks out as typewriter-like, which I think is a huge improvement. (It is true it was only used in headers, but visually the headers dominate the impression the page makes, because they have such large font sizes.)

Presented in the comment the full-width underlining struck me as a bit weird, but I must admit that in the typical article with an Introduction / Setup / Pieces / Rules / Notes section it works.

Having headers in italics still strikes me as odd, though. But having 6 header levels seems overly generous; I was nottt even aware that HTML supported h5 and h6, and I cannot imagine I would ever need more than 3. So I suppose there isn't any real downside to having some of the lower-level headers in an uncommon style; most people won't need them, and the few that do need more than 3 levels but less than 6 can skip the styles they don't like.

The h5 style makes the impression of being 'incomplete', because it doesn't have left and right vertical borders. Even then, it is still a very exotic header style. The typeface of h1-h4 'blends in' very well with the typeface of the main text, but h5 and h6 somehow don't.

There still is a lare size gap between h4 and the main text. I would expect there to exist headers that had nearly the same size as the main text, but distinguish themselves by being boldface.

The font that is now on HTML buttons still seems fixed space. I don' think that is a good idea; it makes the buttons unnecessarily wide.

The h5 style makes the impression of being 'incomplete', because it doesn't have left and right vertical borders.

It also doesn't have horizontal borders. If you are seeing it with any borders, you need to clear your cache.

The typeface of h1-h4 'blends in' very well with the typeface of the main text, but h5 and h6 somehow don't.

Also, those no longer use small-caps. Clear your cache to see how it currently looks.

You might want to read the later comments before replying to earlier comments that are no longer pertinent. If you still see the double underlines, you need to clear your cache.

You are correct - my appologies.  I'm travelling and on my father's computer, so I didn't want to clear cache, but I've opened a different browser so I can see how it looks now.  This is a definite improvement.  I think the H2 text size is very, very large.  I'd like to see that scaled down a bit, but things are looking better.  

You can press ctrl+f5 to clear cache for current page only

I think the H2 text size is very, very large. I'd like to see that scaled down a bit, but things are looking better.

Is that because you are used to it being smaller, or because there are specific pages on which the H2 heading is too long?

I looked into the default sizes of H1-H6, and I learned that by default, H4 is the same size as the body text, and H5 and H6 are even smaller. This doesn't make sense to me. Headings should always be larger than the body text. Additionally, higher level headings should be larger than lower level headings, and size differences between headings of the same style should be large enough to tell them apart. Bearing these considerations in mind results in the headings all being larger than they are by default. This doesn't leave a lot of room for adjusting the size of H2, which is already closer to the size of H3 than to the size of H1.

Additionally, text is appearing larger on Chrome than on Firefox. I normally develop on Firefox, but if you're using Chrome, you will see larger fonts than I normally do. When I inspect the page h1-h6-test.html, which has no CSS or styling, it gives me the same values for the font sizes even though it appears larger on Chrome. In inspecting, Xiangqi, I get the same values for the style elements, though I cannot get Firefox to give me the size in pixels. I need to look into why text is larger on Chrome and see if there is some way to standardize the size in different browsers on the same computer. You may inspect these pages yourself with Developer Tools on Chrome or Web Developer Tools on Firefox.

There are pages on which I plan to use a different style or a lower heading level. These would include link pages, Zillions pages, and Game Courier preset pages, which normally have fewer sections and less text than other pages. I'll work on that later.

I looked into the default sizes of H1-H6, and I learned that by default, H4 is the same size as the body text, and H5 and H6 are even smaller. This doesn't make sense to me. Headings should always be larger than the body text.

Someone might want to add paragraphs in a smaller-than-standard font. And these might need headers too.

Four header levels already seems quite generous to me.

Someone might want to add paragraphs in a smaller-than-standard font.

If you want small text, you can enclose your text in <small>small tags</small>.

But can you also do that with headers?

Instead of setting the body text at 18px, I have erased code for setting the size of the body text, so that it will now use the system default. On the Windows desktop, at least, this is typically 16px. I checked multiple devices, and on each one, it seemed to make the text smaller, though Android and iOS devices did not let me inspect the elements and determine the precise size.

One reason for making this change is that Literata, which is now the body font instead of Lora, has a clearer look at smaller sizes. Another is that it should make the display more responsive to particular devices. Instead of using a fixed size across all devices, it lets the device determine the size. Also, I had previously set the font-size for comments to 16px, and I haven't had trouble reading comments. However, comments will now display text at the same size as the main content instead of at a smaller size. So, comments should generally display at the same size as they used to. Finally, the font size will now better match whatever zoom level someone prefers to use his browser at.

Since the heading sizes are defined with the em unit, which is determined by the body font size, they adjust according to the body font size. So, with the body font a bit smaller on most devices, the headings will also be a bit smaller.

I have adjusted the em values for the headings so that specific integer pixel values will be used when the body font is 16px. These are 36px for H1, 32 for H2, 29 for H3, 26 for H4, 21 for H5, and 18 for H6. Instead of making the differences the same between each pair of neighboring headings, I used smaller differences where the style changed more, and I put a distance of .5em (8px) between distant headings using the same style.