Is Your Lens Properly Assembled?

... or a cautionary tale for DIY lens cleaners.

While it's unlikely many of us would even think about taking a modern lens apart, old plate camera lenses, particularly brass-bodied Petzvals are often filthy inside, plus they have those inviting serrated edges meaning you don't need tools to unscrew them and get at the glass.  It's usually no problem, provided you're careful and make an exact note of where and how everything fits together. This last part is essential - as I'll show...
(this does get a bit nerdy in places but stick with me)

My 'workhorse' camera for wet-plate work is a Watson studio whole plate from about 1917.  I bought it with a nice looking lens; a Cooke, no less bearing the name "No. 2 Cabinet Lens".

This Cooke lens is a typical brass-bodied portrait Petzval with built in lens hood.

Lovely hand engraving 

 

It's a Petzval design: about 8inches (200mm) focal length and about f/4.  The "Cabinet" designation suggests it should cover an area at least as big as a cabinet size carte de visite: around 6.5 x 4.25 inches. (this is 'half plate' size) 

This advertisement for Ross lenses shows something very similar:

The Petzval design is a compromise, giving good sharpness in the centre of the image with lots of brightness at the expense of quite a bit of softness at the edges. This is fine for most portraits and the 'swirly' bokeh effect the design gives is now highly prized.  I've always been a bit disappointed though, the edge falloff is very marked and unless it suits your subject the relatively small 'sweet spot' of sharp focus can be annoying.  It sure doesn't cover a cabinet size!

Collodion test picture (whole plate)  taken with the Cooke Cabinet lens: The fuzzy, swirly effect looks nice here but surely this lens shouldn't be this wild?

The Davidson Lens Measure

A recent addition to my collection of odd photographic gadgets is a Davidson Lens Measure. This has a clock-like dial and three pins: the two outer ones are fixed, while the centre one is connected to the indicator needle. On a flat surface all three pins are in line and the dial registers zero. If the surface is curved, the centre pin will be out of alignment. The gauge therefore shows just how strong a curve, concave or convex a surface has. It's used for measuring the curvature, and hence the strength of lenses.  Mine was cheap as it has a cracked glass, the needle is too short and it's not properly zeroed. I'll fix all of those things - (maybe!) one day but for now it gave me the idea to test my Cooke lens. If it had been cleaned and wrongly reassembled at some time this might be why it's now not as good as I expected.

As the Ross drawing above shows, the Petzval design has four elements: the front two are cemented together to form a plano-convex (or very nearly) shape which would be hard to reverse by accident. The two sides look very different and the rear flat surface would look very odd if it was facing forward. 

The rear set is different: there's a bi-convex element, an air space and a concave-convex element. There are several different ways these can be assembled and only one is correct.

Thomas Sutton’s “A Dictionary of Photography” from 1858 includes this in his description of the Petzval design:

The posterior lens is composed of two lenses separated by a small space; that next the front lens is of flint glass, convex-concave, and divergent, being thinner in the middle than the edges; the other is biconvex, and of crown glass, being placed with its most convex side next to the concavity of the flint…

This tallies with the Ross drawing above.  So how were my lens's elements arranged?

Cooke lens with rear elements, brass holders and spacers and the measure. The black thing is a Waterhouse stop for f/22.

Er.. certainly not in that order!  You can see my rough sketches in the picture above.  This lens had the biconvex first, then the concave-convex.  Easy enough to swap them over but according to Sutton, the biconvex isn't symmetrical. It bulges slightly more on one side than the other. This is hard to see with the naked eye but an easy job with a lens measure:

Measuring the lens curvature. The gauge isn't properly zeroed (it's actually registering this as concave!) but all I need here are the relative curvatures. It reads more on one side than the other so I know which side is which.

The measure made it easy to find the more convex side and turn it to "face the concavity" of the other element. A good clean of all the elements and everything went back together in the right order. 

Testing:

As a simple test I mounted the lens on my LSC 10x8" and loaded some darkslides with RC paper. Here's one made with the lens as acquired:

A: as found. Oh dear! sharpish in the centre but definition falls off fast.

-And here it is with the lens elements in the 'proper order:

B: Elements in the proper order as per Thomas Sutton's 1898 description.  MUCH better sharpness extending almost to the edge.

Just for comparison, I tried reversing the biconvex element (most convex side facing away from the other element):

C: Elements in the same order as B but with the rearmost element reversed. Only a tiny difference in the look of the lenses but a big difference in quality. It's almost as bad as the first one.

Lastly I tried something which works on some but by no means all lenses; removing the rear element set altogether:

D: Both rear elements removed.  The focal length increases as does the coverage - nearly all the way to the edges of the 10x8" sheet.

The effective focal length increased to about 11 inches and the quality is sharp and with good contrast all over. - worth remembering. Ansel Adams was very fond of his Cooke "triple convertible" lens which worked as different focal lengths depending on which elements were used and I have a Schneider Symmar 150mm that becomes a 265mm if you remove the rear set.  This doesn't always work but this shows it's worth trying!

What have I learned?

Firstly, don't take it for granted that an old lens is correctly assembled. When the parts are taken apart for cleaning it's very easy to put something back the wrong way round. Old Victorian lenses may have been apart several times in their lives.  This one even had signs of old modification work (drill holes etc.) in the brasswork at the back so it may have been mis-assembled for a very long time.

Secondly, by looking at the lens design drawings (or the closest you can find) it's possible to work out how the parts should go. Subtle differences in lens curvature may be hard to spot but unless an element is perfectly symmetrical it will make a difference. You don't need a lens curvature gauge (fun gadget though it is): look at the reflections in the lens surface:

 

Here are the two sides of the biconvex element.  The more curved side is on top in the right hand picture as the window reflection is smaller.  Yes, that is a horrendous chip out of the edge! - It's the cause of the chunk out of the image circles in the test images above.  Some people advocate painting such chips black to reduce flare.

Take a look at the edges of the elements, especially thick ones. Often the makers pencilled names or dates there which can identify otherwise unmarked optics.

Lastly, removing the rear elements can give you another longer focal length option. - OR if you really want the 'swirlies' you could deliberately move the elements around. This is what people do with lenses like the Helios-44 but that's a subject for another article.

Hope you enjoyed this. A bit of a nerdy post this time but it might just help you transform that clunky old lens back into the gem it deserves to be!