Andrew's First Touring Bicycle

I got "into" cycling in 1973, and bought my first "real" bike, with derailleur gears and dropped handlebars, in 1974, just before I went up to University. Then, in 1975, after it became clear that I could actually live on the student maintenance grant that the British government then paid to students, I decided to spend the earnings from my summer job on a "dream bike", custom made for my purposes. Actually, I blew most of the money on the frame, and scrimped on some of the components, on the (sound) theory that I could always upgrade the components later as they wore out.

The Harry Perry Touring Bike

I'm still riding this frame today, so I guess that I made some good choices. Stuff that's still original:

• Reynolds' 531 double-butted frame, with braze-ons for:
  • mudguards (fenders to those of you in the USA)
  • rear carrier
  • Mafac cantilever brakes
  • pup pegs on the left rear seatstay
  • Double simplex down-tube gear shift levers
  • Note that water bottle braze-ons were not an option that had ever been thought of at the time, and were not included!
• Cinelli Handlebars and TTT stem
• StrongLight cotterless cranks with TA "CycloTouriste" chainrings
• Brooks B17 Narrow leather saddle

The original setup had a Sturmey Archer 32 hole Dynohub in the front, and a 40 hole Normandy hub in the rear, both built into Weinmann alloy rims. (A Dynohub has a generator built-in; it's heavy, but does not have the friction or slippage problems of a tire-driven generator).

The Original Gearing

Apart from the frame dimensions, the thing that made this bike uniquely mine was the gearing. The shop that built it up for me, Harry Perry Cycles in Woolwich, S.E. London, dealt primarily with racing cyclists, and when I told Paul Mepham, the then owner, what I wanted, his immediate reaction was to ask "why would anyone want gears that low?" My answer was "to go up mountains with camping gear", and indeed, in the summer of 1976, as part of a CTC tour lead by the redoubtable Peter Knottley, I climbed Mt Ventoux in Provence, France, with full camping gear. (I did not go as far as one member of our group, however: he carried up a whole watermelon.)

So what was the manic gearing? First I'll tell you some of the principles.

I wanted wide ratios;

I didn't race, but I did long tours, either Youth Hoteling or camping, in hilly country. The reason to invest in a double chainring at all was to get some lower gears, while keeping the high ones.

 

Gears should be in a geometric progression.

In other words, there should be the same percentage difference between any pair of adjacent gears. This seems obvious now, but at the time there had been a long correspondence in the cycling press between those who thought that even gears meant arithmetic progression, i.e., a constant number of teeth difference in the cogs (like 13 15 17 19 21) and those who thought (correctly!) that it meant a constant multiplier,

 

Chainrings and cogs should have a whole number of teeth!

This is an obvious, but inconvenient, consequence of the technology! It's inconvenient because it limits the degree to which one can achieve a geometric progression.

 

Don't plan to use the "crossed" combinations

That is, don't plan on running the chain from the small ring to the small cog or the large ring to the large cog. These combinations cause excessive wear, even if they can be made to work.

 

The "geometrically spaced ratios" criterion gave me two options:

1. Use the widest block that one could find, and two rings quite close together, so that the gears on the smaller ring fit between the gears on the larger ring at every step. This is sometimes called "half step gearing". For example:
   
   

   Chainrings	48	43		ring ratio		1.116
   small sprkt	14			wheel size		27
   Nr sprockets	5			sprkt ratio		1.243	=	1.1152
   sprkts ideal	13	16.15	20.07	24.95	31
   sprkts integer	13	16	20	25	31
   delta squared	0.0%	2.4%	0.6%	0.3%	0.0%			error =		0.18
   gears	99.7		81.0		64.8		51.8		41.8
   		89.3		72.6		58.1		46.4		37.5
   ideal	99.7	89.4	80.2	71.9	64.5	57.9	51.9	46.6	41.8	37.5
   delta squared	0.0%	1.2%	64.0%	39.6%	7.9%	3.3%	0.4%	1.3%	0.3%	0.0%		error =		1.086

   
   
   Let me try to explain this chart.

   The Yellow boxes are the inputs: 48-43 chainrings, and 13-31 block. Why these particular sizes? Well, 13-31 was about as wide a ratio block as one could find at the time. And the combination 13 16 20 25 31 happens to be very close to a geometric progression. The row labeled "sprkts ideal" gives the number of teeth that one would hypothetically need to get a geometric progression with inter-sprocket ratio of (31/13) ^1/4 =1.243. The row labeled "delta squared" gives the square of the difference between the gears that one gets with the listed real sprockets (all of which have a whole number of teeth) and the ideal. The sum of the deltas is 0.18, which is about as good as one can get.

   To space the gears from the two chainrings so that they intersperse, the ratio of the chainring sizes should be the square root of the ratio of the sprocket sizes, or 1.115. We find that 48-43T rings match this desired gap very closely, and produce the gearing (in inches) shown in the blue boxes. The five gears in the first blue row are obtained on the outer chainring; the five gears on the second blue row are obtained on the inner chainring. The delta squared row shows the square of the deviation from perfect geometric ratios.

   The deficiencies of this scheme (which was quite common at the time) are pretty obvious:

   1. It does not satisfy the "wide ratios" requirement.
   2. If you want to avoid the crossed combinations, you don't actually get to use the 89.3" or the 41.8", so you don't get even gearing at all.
   3. Every other gear change is a double change, e.g., while going from 81.0" to 72.6" can be done on the front changer only, going down the next step to 64.8" means shifting down at the back while simultaneously shifting up at the front.

    

2. The second option was to put all the high gears on the big ring , and all the low gears on the small ring. I called this "Land Rover gearing" because, like a Land Rover, it gave the bike distinct high and low ratios. Once again, I searched for a combination of sprockets that were close to geometric progression, and found those in the following chart.
   
   

   rings	50	30		ring ratio		1.6666
   small sprkt	14			wheel size		27
   Nr sprockets	5			sprkt ratio		1.1996
   sprkts exact	14	16.79	20.15	24.17	29
   sprkts integer	14	17	20	24	29
   delta squared	0.0%	4.2%	2.2%	3.0%	0.0%			error =		0.31
   gears	96.4	79.4	67.5	56.3	46.6
   				57.9	47.6	40.5	33.8	27.9

   
   
   The idea here is that one uses only the four small sprockets on the outer ring (and thus one doesn't actually use the 46.6" gear) and only only the four large sprockets on the inner ring (excluding the 57.9" gear). Because if this, the gap between the chainrings is chosen to be the the cube of the gap between the sprockets, giving the close-to-geometric set of 8 gears shown in blue. There is only one double change required, when one moves from 56.3" to 47.6", and in practice the most crossed ratios work well enough that it was ok to change into them briefly, realize that the chain was grinding, and then make the double change.

   This option met all of my criteria.

Not surprisingly, the enormous chainring gap, going from 30T to 50T, was greeted with some skepticism by the experts, who told me that it would never work. But when we built up the bike, we found that it worked just fine, even with the cheapest and simplest front changer (which I already had in my "parts box"), a Simplex. (I later replaced this with a Shimano 600 changer,which looked much nicer, and worked just as well.) The rear changer turned out to be critical, since it had to deal with a lot of slack chain: from (50+24) to (30+17), or 27 teeth. The longest arm rear changer available at the time was top-of-the-line Shimano Crane, which I could not afford. The cheaper long-arm Titlelist worked just fine, however.

At the time, rear freewheel blocks used interchangeable sprockets, so building-up the one I wanted posed no special problems, except that the 29T large sprocket had to be special ordered; 28T was the more common option. But the largest sprocket never wears out, so I was able to reuse it several times, even though I replaced the rest of the block. It was originally a Regina, and at some time when (I became more wealthy) it was upgraded to a Regina Oro.

The Twentieth Anniversary Overhaul

This gearing was essentially unchanged, except for maintenance replacements, until the bike's 20 birthday, in 1996. I had noticed that the power train was not as quiet as it had once been, and also that I had picked up a bulge in the rear rim, which rendered the rear brake useless. Since this meant a new rim and a wheel rebuild anyway, I determined to replace the rear hub with a freehub, which is mechanically superior to the old freewheel and block. Since freehubs came with 7-speed clusters, this meant refiguring the grearing. Mark at Northwest Bicycles did all of the work for me, retaining most of the integrity of the bike while taking advantage of the new technology of a Hyperglide cluster. Here is the current gearing:

rings	50	30		ring ratio		1.66
small sprkt	14			wheel size		27
overlap	3.29
Nr sprokets	7			desired sprkt ratio		1.148		actual mean		1.148
sprkts ideal	14	16.07	18.44	21.17	24.29	27.88	32.00
sprkts integer	14	16	18	21	24	28	32
delta squared	0.0%	0.5%	19.5%	2.8%	8.6%	1.4%	0.0%	error =		0.57
gears	96.4	84.4	75.0	64.3	56.3	48.2	42.2
					57.9	50.6	45.0	38.6	33.8	28.9	25.3

With a Hyperglide rear cluster one can't select the sprockets individually; one has to choose one of the prepackaged selections that Shimano sells. But the 14-32 is actually pretty close to geometric; it looks like it was designed by someone who understood the concepts of gearing!

The seven speed block means that the gears are closer together, and the overlap between the high and low ratios is now three gears. Once again, avoiding the most crossed pairings is a good plan, but that still give us two ways of obtaining a gear around 49": the 48.2" or 50.6" combinations. Actually, with this cassette, the rings should be a little more widely spaced; 50-29T would work nicely, but we had a lot of trouble finding a replacement for the 50T chainring (which was worn out), and decided to reuse the existing 30T. This worked beautifully, with a Deore XL rear changer which has a nice long take-up arm. This is the bike that I rode on Cycle Oregon XI, and which climbed Larch Mountain in July 1999. Its overall weight is about 28lbs. Andrew P. Black Last modified on 1999.08.11 at 16:09