Ferric chloride and oxalic acid 100
Ferric oxalate 89
Ammonium ferric oxalate 80
..
Ammonium ferric citrate 15
Obviously cyanotype (based on the citrate) is a lot less sensitive than
palladium (presumably using the oxalate). It's interesting to note that
according to Cassell's Cyclopedia ferric ammonium oxalate (used more now
for printing-out platinum) was then "used occasionally for blue prints"
(the equivalent of a hotted up or supercharged cyanotype?).
The densitiometer I could normally use is broken and will probably take
two months to fix so I can't provide "cyantific" rows of numbers. Here
are a few impressions from two series of tests, all with a single bulb
at 10 minutes exposure (same stepwedge, same printing frame, emulsions
all sensitized on the same paper):
Argyrotype (all with printing frame in direct contact with tube)
Philips TLK 40W/03 - darkest Dmax - 10 steps
NEC 20W BL - 11 steps more clearly defined (better highlight separation)
Philips TLK 40W in 20 watt fitting - almost identical to above but with
one less highlight step.
Thorn 18W Cool white - 8 steps, very anaemic Dmax
Philips TLD 18W/33 White - virtally identical to above.
Cyanotype (varying distances): Philips TLD 18W/33 White in direct contact
with the tube was found to be virtually identical to Philips TLK 40W UV
tube at 10 cm. Number of steps visible hardly changes with distance
although overall density drops off horrendously. Presumably if I extended
the time with the normal fluorescent I could produce densities equivalent
to to the BL tube, but I gave this up as a waste of time and electricity.
Why the 20W BL should perform best overall seems a bit of a mystery.
Changing the subject from the ferric processes to carbon, I couldn't
understand Sandy King's results showing GE F20 T12D (Daylight) tubes
"were both faster (if we measure sensitivity by the time required to
produce the first maximum black on steps 1 and 2), and of greater
contrast than the UVs."
Then Mike Ware's comment about the *area* under the curve of a lamp's
spectral comment clicked. To restate it: Sandy's daylight tubes are
actually delivering more useful actinic light than his 350 nm UVs, even
though the UV tube has its maximum output at a wavelength closer to the
commonly cited optimum for dichromated colloid. Imagine a graph of the
spectral output of a specific tube superimposed on top of a graph giving
the spectral sensitivity of the dichromated colloid:
|
| * Peak at 367 (ammon) and 357 (potassium dichromate)
| * * *
| * * *
| * *
|_______________________*
300 350 400 580 Sensitivity falls to zero
Wavelength (nm) around green yellow
(Sorry about the graph - see Kosar p. 74 for the original)
The spectral curve of Sandy's daylight tubes must cover a much larger
overlapping *area* than his UV tubes, thereby giving a much greater total
exposure. That explains why the daylight tubes were "faster"; regarding
contrast it's also important to remember that absorption of light varies
with wavelength. The shorter rays of an ultraviolet rich source "are
scattered and absorbed in the top layer of the photoemulsion, [while] rays
of longer wavelength penetrate more deeply, and produce prints of higher
contrast (or in other words, a thicker insolubilized layer)" (Kosar p.
97). Forget quantum yield, maybe this could be likened to a 450 nm ray
dropped from the same distance making a much deeper dent in the gelatine
than an otherwise equivalent 350 nm ray. However above 450 nm, spectral
sensitivity appears to be falling off rapidly; thus as Eder's results
(cited by Klaus Pollmeier) suggest maybe the optimum wavelength is between
400 and 450 nm.
As Klaus and Bob Schramm have pointed out glass transmittance is highly
relevant. Klaus said glass just begins to let the light through at around
350nm; Kosar says the cut-off point is 340 nm; Crawford says 300 nm. Given
the worst case scenario it's easy to see why say the Philips TL 40W/03
super actinic (with a peak at 420 nm) might perform better than their
TL/05 actinic series emitting UV radiation between 300 and 460 nm with a
maximum at 365 nm - the left hand side of the /05's spectral curve is
rather abruptly chopped off:
Glass| Useful
cut-off| actinic light
| *
| * *
| * # *
| * # *
|__*_____#_____________*_
300 ^ 365 460
Maybe with normal (rather than quartz glass) daylight tubes do have some
applicability in regulating contrast for dichromated colloid processes.
Kosar, p. 98 gives a graph showing characteristic curves (actually very
straight lines) for monochromatic light of different wavelengths; he also
says radiation containing multiple wavelengths produces a slightly curved
rather than a straight line. If the "internal filter effect" wasn't bad
enough, here's yet another variable. Back to those tests, Judy!
Sorry this is so long winded; I've really just re-stated points made by
others, but writing it helped me understand the factors involved a bit
better.
Philip Jackson
pjackson@nla.gov.au