Contrast Filtration and Split-Grade Printing

A filter number is not a contrast setting. It is an instruction about the colour of the printing light, and the paper turns that colour into a contrast because of how it was built. Once that is clear, three things a printer does every session stop being habits and become arithmetic: why a test strip made at one grade sometimes survives a change of filter and sometimes does not; what two exposures through two filters actually add up to on one sheet; and which of the many claims made for split-grade printing the published evidence will carry.

Part XVIII owns the material — three emulsion components, one blue speed, three green speeds — and that page is not repeated here. This one is written from the easel: it takes the maker’s published numbers and turns them into decisions you can record.

Choosing the grade twice

The grade gets chosen twice in a printing session, by two different methods, and the second choice is allowed to overrule the first.

The first choice is a measurement. Reading the negative produced an effective density range for the projected image. ILFORD’s instruction is to multiply that figure by 100 and take the nearest published range figure — a negative measuring 1.32 log exposure units gives 132, and on MULTIGRADE RC DELUXE the nearest published figure is 130, which is filter 0. The logic is simply that the paper’s exposure scale should be about as long as the negative’s density range, so that neither end of the negative falls off the end of the paper.

The second choice is a judgement about the picture, and Kodak state the rule for it in one sentence: if the strip looks flat or muddy, use a higher-numbered filter; if it has a harsh, contrasty appearance, use a lower-numbered one. That is not a correction of the measurement. It is a different question being asked.

The measurement answers what will fit. The judgement answers what the print is for. A landscape whose whole subject is the separation between two nearly identical greys wants more contrast than the fit requires, and will pay for it by losing the top of the sky and the bottom of the wood. A portrait with a specular highlight on the forehead may want half a grade less than the fit allows, so that the highlight has somewhere to sit. Neither is a mistake, and both are choices you have to be able to defend and record.

📷The two questions are separable, and printers who confuse them get stuck

Ask them in order and they stay apart.

Question one, arithmetic: does the negative’s range fit the grade’s range? A test strip read as a measurement answers it — if the highlight lands two bands below the shadow, the negative is flatter than the grade and wants a harder one, and if the shadow blacks before the highlight has any tone, it wants a softer one.

Question two, intention: given that it fits, is the fit what the picture needs? Only the two sentences you wrote before you started can answer that.

The stuck printer is the one who has answered question two by changing the grade and then finds the ends of the scale falling off, and answers that by changing the grade back. The way out is to say which question the change is answering, and to accept the cost of the answer explicitly on the map.

A negative that measures outside the published range is a third case. If the effective density range comes out at 1.85 and the softest filter on your paper is R170, the paper cannot hold the negative, and no choice of filter changes that. What changes it is the negative — reprinting from a different one, or accepting that one end will be empty and deciding which end. That decision belongs on the map too.

What the filter selects, in the printer’s terms

Part XVIII’s mechanism, restated as an instruction rather than as chemistry.

The emulsion is three separate emulsions mixed together. All three have the same inherent contrast and the same speed to blue light; they differ only in how much green sensitising dye each carries, and therefore in their speed to green. ILFORD’s consequence follows immediately: blue light makes all three react together over a narrow exposure range, which is high contrast, while green light is answered first by the most heavily dyed part alone, over a much wider range, which is low contrast.

So a filter does not change the paper’s contrast. It changes which components are doing the work, and the components are what have the curves.

• Magenta absorbs green and transmits blue. Blue is the hard direction.
• Yellow absorbs blue and transmits green. Yellow is the soft direction.
• Every mixture between them gives a partial stagger of the three curves, which is why the grades are a continuum and a maker can sell twelve filters and once sold a head resolving tenths.

What each of the two exposures delivers to each of the three components

1. 1Equal segments: the hard exposure — all three components share one blue speed, so a blue exposure adds the same amount to each
2. 2Unequal segments: the soft exposure — the three green speeds differ, so a green exposure adds most to the most heavily dyed component
3. 3The boundary between them — move it left or right at constant total length and you have changed the grade without changing the density

Drawn to teach. No maker publishes the individual components' speeds, so the segment lengths show the relation ILFORD describe and are not measured from a material.

The number on the filter is not a property of the paper. ILFORD say it themselves about colour heads: the published filtration tables “can only be a guide, because individual enlargers vary”, and the actual filtration for a particular enlarger “must be determined by trial”. The transferable quantity is the range figure, which is a measurement of the paper’s exposure scale under a stated system; Part XIII owns it.

Exposure factors across the filter range

Here is where a mechanism becomes a number you can be caught out by. Changing filter changes how much of the lamp’s output reaches the paper, so it changes the exposure — and how much depends on the paper as much as on the filter.

ILFORD publish an ISO speed for every filter on every current MULTIGRADE paper. The ratio of two speeds is the filter factor between them, and its logarithm to base 2 is the change in stops.

stops = log₂(P_softer ÷ P_harder)

A published speed pair, converted to stops

P is the ISO paper speed at a given filter, the two values being read from the maker’s own table for one product. A ratio of 2 is one stop; a ratio of 1.09 is about an eighth of a stop.

ILFORD paper	ISO P, filters 00–3	ISO P, filters 4–5	Factor	In stops
MULTIGRADE RC DELUXE (current)	240	220	1.09	0.13
MULTIGRADE RC PORTFOLIO (current)	240	220	1.09	0.13
MULTIGRADE FB CLASSIC	230	210	1.10	0.13
MULTIGRADE FB COOLTONE	250	225	1.11	0.15
MULTIGRADE IV RC DELUXE	200	100	2.00	1.00
MULTIGRADE RC COOLTONE	200	100	2.00	1.00
MULTIGRADE RC WARMTONE	100	50	2.00	1.00
MULTIGRADE FB WARMTONE	100	50	2.00	1.00

The speeds are ILFORD’s; the factors and the stop column are the course’s arithmetic on them. Two things fall out at once.

The step is at the same place on every product. The speed is flat from filter 00 to filter 3 and steps between 3½ and 4, exactly where the filter leaflet says the speed matching ends. Nothing surprising happens inside 00 to 3½ on an ILFORD paper with ILFORD filters, and that is the whole content of the phrase “speed matched”.

The size of the step is not the same on every product, and it differs by a factor of nearly eight across one maker’s own range. On the warm-tone papers it is a clean stop. On MULTIGRADE FB CLASSIC and the current RC DELUXE the published speeds imply about an eighth of a stop.

🧮Three ILFORD statements about the same step, and what the tables do to the disagreement

Part XVIII flagged this as an unresolved disagreement between two ILFORD documents. The full set of tables narrows it, though it does not close it.

• The contrast control sheet: the exposure time for filters 00 to 3½ is the same, and that for filters 4 to 5 is double.
• The filter leaflet: grades 4 to 5 will generally require more exposure, “in practice this will be between 0.5 – 1 stop depending on the product”.
• The FB CLASSIC and FB COOLTONE sheets: filters 4 to 5 need “typically between x1.5 to double the time”.
• The ISO speed tables: a full stop on four products, and about an eighth of a stop on four others.

Three points can be made from that, and the fourth cannot.

First, the leaflet’s “depending on the product” is doing real work. On MULTIGRADE IV RC DELUXE and on both warm-tone papers, the speed table’s step is exactly the factor of two the contrast-control sheet states. Those documents agree.

Second, the current RC DELUXE is a redesign. ILFORD’s own comparison sheet says the new paper “no longer has a large step change in exposure between grade 3.5 and grade 4”, so that only fine tuning is required — which is precisely what P240 against P220 says.

Third, on FB CLASSIC the two ILFORD figures still disagree, and by a lot: P230 against P210 is 0.13 of a stop, and the same sheet’s prose asks for 1.5 to 2 times, which is 0.6 to 1 stop. They are not measuring the same thing. An ISO speed is defined at one density point under a specified procedure; a practical filter factor is what a printer must do to keep a whole print looking the same, and a speed point near the foot of the curve can move very little while the exposure that places the whole print moves a great deal.

What cannot be said is which of the two is right for your box. Measure it: print a strip at filter 3, print the same negative at filter 4½, and record the factor in stops on the inside of the filter drawer.

Foma publish the factor directly, and it is not always flat

Foma print a lengthening factor column beside the speed and range figures, which is the same quantity stated outright. On FOMABROM VARIANT with ILFORD filters it reads 2.4 at filters 00 to 3 and 4.8 at 4 and 5, with the note that exposure for 0 to 3 is the same and for 4 to 5 should be doubled — the flat run and the doubling, agreeing with ILFORD’s account.

The factors are stated relative to the unfiltered exposure, and the arithmetic confirms it. Foma’s own Variant filter table for that paper gives speeds of 310, 360, 500, 360, 240, 190 and 110 against factors of 1.6, 1.4, none, 1.4, 2.1, 2.6 and 4.6. Dividing the unfiltered P500 by each of the six filtered speeds gives 1.61, 1.39, 1.39, 2.08, 2.63 and 4.55 — the published column, to the last digit. A lengthening factor is the reciprocal of the speed ratio, which is worth knowing because it means the ILFORD speed table above can be read as a factor table without any further assumption.

Then there is FOMATONE MG Classic, which breaks the habit the ILFORD numbers teach.

Published exposure factors against the paper's own range figure

• FOMABROM VARIANT, ILFORD filters
• FOMATONE MG Classic, ILFORD filters
• FOMATONE MG Classic, Foma Variant filters

Show the numbers behind this plot

Three sets of published points. FOMABROM VARIANT under ILFORD filters is a step: the factor sits flat at 2.4 from range figure 160 down to 70, then jumps to 4.8 at range figures 60 and 50, which are filters 4 and 5. FOMATONE MG Classic under the same ILFORD filters is a U: 3.4 at range figures 140 and 120, 3.0 at 105, a minimum of 2.4 at 90 and 2.0 at 75, then back up to 2.4 at 70 and 3.0 at 55. FOMATONE under Foma's own Variant filters is a shallower U with the same shape, from 2.0 at range figure 120 down to a minimum of 1.5 at 105 and 80, then 1.8 at 75, 2.0 at 65 and 3.0 at 55. The lesson of the picture is that the flat, speed-matched middle that ILFORD's papers show is a property of a particular paper and filter set together, not of variable-contrast printing, and that on FOMATONE the exposure changes by three-quarters of a stop inside the range a printer would expect to be flat.

Series	ISO(R) range figure — soft to the right, hard to the left	Published lengthening factor (× unfiltered)
FOMABROM VARIANT, ILFORD filters	50.00	4.80
FOMABROM VARIANT, ILFORD filters	60.00	4.80
FOMABROM VARIANT, ILFORD filters	70.00	2.40
FOMABROM VARIANT, ILFORD filters	90.00	2.40
FOMABROM VARIANT, ILFORD filters	110.00	2.40
FOMABROM VARIANT, ILFORD filters	130.00	2.40
FOMABROM VARIANT, ILFORD filters	160.00	2.40
FOMATONE MG Classic, ILFORD filters	55.00	3.00
FOMATONE MG Classic, ILFORD filters	70.00	2.40
FOMATONE MG Classic, ILFORD filters	75.00	2.00
FOMATONE MG Classic, ILFORD filters	90.00	2.40
FOMATONE MG Classic, ILFORD filters	105.00	3.00
FOMATONE MG Classic, ILFORD filters	120.00	3.40
FOMATONE MG Classic, ILFORD filters	140.00	3.40
FOMATONE MG Classic, Foma Variant filters	55.00	3.00
FOMATONE MG Classic, Foma Variant filters	65.00	2.00
FOMATONE MG Classic, Foma Variant filters	75.00	1.80
FOMATONE MG Classic, Foma Variant filters	80.00	1.50
FOMATONE MG Classic, Foma Variant filters	105.00	1.50
FOMATONE MG Classic, Foma Variant filters	120.00	2.00

Measured, in the sense that every point is a number Foma publish in the technical data table of the paper's own sheet. The two axes are the sheet's own ISO R column and its own lengthening-factor column; nothing has been interpolated.

On FOMATONE with ILFORD filters the factor runs 3.4, 3.4, 3.0, 2.4, 2.0, 2.4, 3.0 across filters 00 to 5. It is not flat and it is not monotonic: the cheapest exposure is at filter 3, and it rises in both directions. Between filter 00 and filter 3 the factor changes by 3.4 ÷ 2.0 = 1.7, which is 0.77 of a stop — inside the range a printer trained on ILFORD papers would expect to cost nothing.

The practical rule, and it is the covering one: speed matching is a property of a paper and a filter set together, and it is a design decision by one manufacturer about their own products. A test strip made at one grade is valid at another only where the maker’s own table says the speed does not move, and only if the paper, the filters and the lamp are the ones the table was written for. Everywhere else, re-anchor. Working in stops rather than seconds makes re-anchoring one addition instead of a recalculation.

✕Two ways the filter costs you exposure that are nothing to do with grade

The filter itself costs about a stop. Every ILFORD sheet gives an unfiltered speed alongside the filtered ones, and the gap is large: P500 unfiltered against P230 at filter 2 on FB CLASSIC is 1.12 stops, and P590 against P250 on FB COOLTONE is 1.24. That is the price of putting anything in the light path at all, and it is why the unfiltered column is the fastest number on the sheet. It has no bearing on the grade, and it is why a printer who removes the filter to focus must remember to put it back.

A dirty filter costs contrast, not just light. ILFORD state that marks from dust, rough handling or finger marks “will reduce print contrast and degrade the image”. Why a fingerprint should cost contrast rather than only sharpness is not explained on the leaflet; the course’s reading is that a scattering filter passes light around its intended spectrum as well as through it, which is a leak in the direction of the unfiltered case, and the unfiltered case on a MULTIGRADE paper is a middle grade. Handle by the edges; store in the envelope; and if a hard grade has quietly gone soft over a year, look at the filter before you blame the paper.

Split-grade printing, described mechanically

Two exposures, one sheet, two filters. Give the paper a timed exposure through a hard filter, change the filter without moving anything, and give a second timed exposure through a soft one. Process once.

Mechanically that is all it is, and three published facts say why the sheet does not mind.

The latent image adds. ILFORD state that no significant change in picture quality is seen when MULTIGRADE RC papers are left for 24 hours between exposure and processing. A pause of ten seconds to change a filter is nothing on that scale, so the second exposure lands on a latent image that is still entirely there.

Nothing about the two exposures is a special case. Each is an ordinary exposure of an ordinary paper through an ordinary filter, obeying the same speed table as any other.

The order does not matter, on the published evidence. Nothing in the corpus distinguishes hard-first from soft-first, and the additivity above gives no reason for a difference. Pick one order and keep it, because a printing map that does not fix the order is a map with a variable in it.

⚗The one classical caveat, and why it does not bite here

Sheppard and Mees devoted a section of their 1907 investigations to the fact that a series of intermittent exposures does not always produce the same effect as the equivalent continuous exposure — first shown by A. and L. Lumière in 1881, then investigated by Abney. Their own measurement ran a sector wheel at 1520 and at 9.5 revolutions per minute for the same total exposure and got two different curves, the difference concentrated at the bottom exposures. Their conclusion for practical sensitometry was that below 100 revolutions per minute the error would be negligible even for slow plates.

Englisch’s explanation, which they adopt, is a fading loss: a reverse reaction that undoes part of the latent image during the dark periods, so the loss grows “the longer the pause, and the more numerous the pauses”.

A split-grade print has exactly one pause. On the number of pauses it is the most favourable case there is; on the length of the pause it is the least favourable, since changing a filter takes seconds rather than milliseconds. The two considerations pull in opposite directions and neither has been measured on modern paper in this corpus — but ILFORD’s 24-hour latent-image statement measures the fading loss from the other end, and finds nothing significant over a period nearly four orders of magnitude longer than a filter change. The course’s position: additivity across a split is treated as sound, on that reasoning, and the reasoning is the course’s rather than a maker’s.

The caveat that does bite is reciprocity, and Foma name it. Their warning for FOMATONE MG Classic is that the Schwarzschild effect shows particularly at exposures above 15 minutes and at the low light intensity characteristic of that paper. A split pair on a slow paper at a small aperture can put one of its two exposures into that region while the other is nowhere near it, and then the two are no longer being metered on the same scale. Open the lens.

Is a split pair the same as the equivalent single grade?

This is the claim the technique is usually sold on, and it needs taking apart carefully, because the answer the mechanism gives is not the answer the printing literature usually gives.

What the mechanism says. Follow ILFORD’s own description to its conclusion. The three components share one blue speed and differ only in green, so the light reaching the paper can be described by two numbers: how much of it the components all respond to alike, and how much of it they respond to differently. Call them the blue quantity and the green quantity. A single filter, for a chosen time, delivers one pair of those quantities. A split pair delivers the sum of two such pairs.

Two exposures adding to a total is arithmetic, so if a single filter exists that delivers the same blue quantity and the same green quantity, the paper cannot tell the difference. The components received the same effective exposures; the combined curve is the same curve, in slope and in shape.

What the practical gain therefore is. Not a new curve. Three things that are real all the same:

1. A continuum instead of twelve steps. A filter set offers twelve fixed blue-to-green ratios. Two times offer every ratio between the two filters used. ILFORD thought that granularity worth productising once — the MULTIGRADE 600 head gave contrast in tenth-grade steps — and a split pair gets the same continuity out of two pieces of plastic.
2. Two exposures that can be treated differently. This is the large one, and it belongs to the next section: a burn can be given entirely through one filter, which no single exposure can imitate.
3. Two handles instead of one. The hard time and the soft time are set independently, so grade and density stop being one dial that moves both.

⁇Where the evidence for the shape claim runs out, in both directions

The claim in circulation is that a split pair produces a combined curve of a different shape from the single grade of equivalent slope — that the technique reaches somewhere filtration alone cannot.

What is established. Two exposures through two filters sum on the paper, because a photographic emulsion integrates exposure and the latent image on these papers is stable for 24 hours. Each exposure can be given its own dodging and burning. Both statements rest on manufacturer literature.

What is not established. That the combined curve differs in shape from the equivalent single filter’s. No manufacturer in this course’s corpus publishes measured characteristic curves for split exposures, and the search was made specifically for them. ILFORD print curves by filter; Foma print sensitometric curves by filter; nobody prints a curve for a hard-plus-soft pair.

Which way the absence cuts. Part XVIII declined to make the shape claim. This page goes one step further and says that the mechanism as the maker describes it predicts no shape difference at all, provided the two exposures are additive and reciprocity holds. So the burden has moved: the shape claim now needs evidence not merely because it is unsourced, but because the published account of the material argues against it. The assumptions that would have to fail for it to be true are named and they are testable — the three-component description is the maker’s simplification, additivity across a pause is inferred rather than measured on paper, and reciprocity is known to fail at long exposures on at least one warm-tone paper.

The measurement that would settle it, for anyone with a step wedge and the densitometer this course builds: expose a step wedge at filter 2 alone and plot the curve; then expose an identical wedge with the hard-and-soft pair your own calibration says is equivalent to filter 2, and plot that. Same paper, same box, same developer, same session, same drying. If the two curves lie on top of one another, the mechanism above is confirmed for that paper. If the split curve has a different toe or a different shoulder, that is a real result, and it is publishable as the course’s own data with its conditions stated.

Nothing on this page depends on the answer. Every use of split-grade printing taught here is a control argument, not a curve argument.

🕰Who invented it: the course does not know, and will not guess

The origin of split-grade printing as a technique, and the person or company credited with it, could not be established. No manufacturer publication, patent, museum or conservation source in this course’s corpus names an originator, a date or a first description, and the attributions that circulate in printing communities are Tier 3 by the course’s own standard and may not be used as evidence.

What can be dated is the material precondition: a paper whose contrast is set by the colour of the light had to exist first, and Part XVIII gives that sequence from the Getty atlas — Fisher’s unmarketed 1911 patent, Defender’s product in 1940, Ilford’s Multigrade reaching the market in 1950. The technique cannot predate them. Beyond that the page makes no attribution, because a history invented to fill a gap is worse than an admitted gap.

The arithmetic, in both directions

A split pair is two times. To move between it and a single grade you need a way of writing those two times that does not change meaning when the enlarger head moves. Here is one, derived from the additivity argument above; it is the course’s own bookkeeping, not a manufacturer’s, and its calibration is empirical.

Anchor each filter on its own black point. The previous page established the anchor: the exposure at which the negative’s clear rebate first reaches maximum black. Measure it twice — once through your hard filter, once through your soft one — and call the two times t_hard and t_soft. They are properties of your paper, your filters and your lamp, and they are the only two numbers in this scheme that have to be measured.

✕Do not predict one anchor from the other with the speed table

The temptation, having just read the speed table, is to measure one black point and calculate the other. It does not work, and the reason is worth having.

On MULTIGRADE FB CLASSIC the published speeds at filter 00 and filter 5 are almost the same — P230 and P210. But an ISO paper speed is defined at one density point, not at maximum black, so equal speeds mean the two curves pass through the same place near the bottom of the scale and say nothing about where they end. The range figures say what happens after that: filter 00 has 1.70 log units of scale to climb and filter 5 has 0.50. If the two curves coincide at the speed point and the soft one is the shallower, the soft one must arrive at maximum black a long way later. Its black point is not a tenth of a stop from the hard filter’s; it is a substantially longer exposure.

How much longer cannot be computed from the published data, because no ILFORD sheet in the corpus says where on the curve the ISO 6846 speed point sits, and the answer depends on exactly that. So both anchors are measured, on the same evening, on the same box, and neither is derived from the other. If yours come out close together, something is wrong with one of the two strips.

h = H ÷ t_hard and s = S ÷ t_soft

Each exposure as a fraction of its own filter's black point

H and S are the two exposure times you actually give, in seconds. h and s are dimensionless: a value of 1 means that exposure alone would have reached maximum black through the rebate.

r = h ÷ (h + s)

The hard fraction, which sets the grade

q = h + s

The total dose, which sets the density

r runs from 0 (soft filter only) to 1 (hard filter only), and under the additivity argument it is the blue-to-green ratio expressed in the printer’s own units. q is how much exposure the sheet received in total, in units of one black point.

The two levers are now separate, which is the whole point. Multiply H and S by the same number and r does not change: the print gets darker or lighter at constant grade. Move H up and S down so that q stays where it is, and the density holds while the grade hardens. That separation is what a single filter cannot give you, because a single filter moves both at once — which is exactly why ILFORD warn that the yellow and magenta filters in a colour head “have not been arranged to equalise exposures”, so that new exposure times have to be recalculated whenever the contrast is changed.

H = r × q × t_hard and S = (1 − r) × q × t_soft

From a chosen grade and density back to two times

🧮Worked both ways, on MULTIGRADE FB CLASSIC

The times below are an illustration of the notation. Yours will be your own, and the whole method depends on that.

Measured anchors. Through filter 5, the rebate first blacks at t_hard = 12.0 s. Through filter 00, at t_soft = 60.0 s — five times as long on this box, for the reason in the block above. The ratio between them is a property of your paper, filters and lamp, and is the one thing here you should not copy from anybody.

Forward: what is this split pair? You printed 3.6 s hard and 42.0 s soft.

• h = 3.6 ÷ 12.0 = 0.30
• s = 42.0 ÷ 60.0 = 0.70
• q = 0.30 + 0.70 = 1.00, and r = 0.30 ÷ 1.00 = 0.30

A hard fraction of 0.30 at exactly one black point — which is where a base exposure anchored on maximum black naturally sits, so q near 1 is the normal case rather than a coincidence. Write r and q on the map beside the two times, because they are the pair that survives a change of enlargement.

Backward: reproduce it half a stop darker. Half a stop is a factor of 2^0.5 = 1.414 on the total, so q becomes 1.414 and r stays at 0.30.

• H = 0.30 × 1.414 × 12.0 = 5.1 s
• S = 0.70 × 1.414 × 60.0 = 59.4 s

Both times are the originals multiplied by 1.414, which is the check: any calculation that changes the ratio when you only asked for density has gone wrong.

Backward from a single grade. Suppose your calibration says filter 2 on this paper corresponds to r = 0.30. Then any single-filter print at filter 2 has its split equivalent at that r, and the only thing left to choose is q — which you get from one test strip on the split pair itself, because the pair’s own black point is where q = 1.

What has to be calibrated, and how little of it there is

The scheme above gives grade as r and density as q. What it does not give is the translation from r to a grade number, and it cannot, because no maker publishes the blue and green content of their filters. That translation is measured once, on your own head, in the same way ILFORD say colour-head filtration must be determined by trial.

Calibrating the hard fraction against the filter numbers you already own

1. 1Measure the two black pointsthe rebate exposure at your hard filter and at your soft one, from the previous page’s method — two numbers, and the only measured ones
2. 2Make a single-filter reference printfilter 2, base exposure anchored the same way, dried and judged dry — this is the print the splits are matched against
3. 3Bracket the ratio, not the timethree split pairs at r = 0.25, 0.40 and 0.55, each with q set to match the reference’s density — one of them will be close
4. 4Repeat once at filter 4two calibration points are enough, because r rises monotonically with the blue share; a third at filter 00 is worth having if you print soft
5. 5Write the table on the filter drawergrade against r, dated, with the paper and the lamp named — it belongs to that chain and not to you

Five sheets and one evening buys a two-way conversion between the way you print and the way every published table is written. The claim that r rises monotonically with grade is the course's inference from the mechanism, not a measurement.

Where split-grade earns its place, and where it does not

It earns its place in local control, and this is the argument that survives everything above. A burn can be given entirely through one filter. Adding 2/3 stop to a sky through the hard filter is not the same move as adding 2/3 stop through the soft one, or through the same mixture as the rest of the print: it darkens the sky and separates the cloud from it, because the extra exposure is going almost entirely to the components that answer to blue. No single-filter print can do that, whatever its grade, because there is only one exposure to modify.

It earns its place in the near-independence of the two ends of the scale, and the reason is in the published range figures rather than in anything mysterious. On FB CLASSIC filter 5 is R50 and filter 00 is R170 — an exposure scale of 1.66 stops against 5.65.

• In the print’s shadows, where the negative is thin and a great deal of light gets through, the hard exposure runs off the end of its short scale almost at once and the tone is at maximum black. The soft exposure contributes there too, but the tone is already as dark as the paper goes.
• In the print’s highlights, where the negative is dense and little light gets through, the hard exposure is still near the foot of its curve and contributing very little. Almost all the tone that exists there came from the soft exposure.

So the hard time mostly moves the blacks and the soft time mostly moves the highlights — approximately, and the approximation is best when the two filters are far apart and worst in the middle of the scale, where both are contributing. Stated as the course’s reading of the published range figures rather than as a manufacturer’s claim.

It does not give extra contrast range. The hardest print available from a split pair is the one where r = 1, which is the hard filter alone; the softest is r = 0, the soft filter alone. Every mixture lies between. On FB CLASSIC that is R50 to R170 either way — the same two bounds the filter set already had. Split-grade interpolates the maker’s range finely; it does not extend it.

Whether it is sharper is not established, and the course makes no claim either way. No measurement of resolution or of edge effect comparing split-grade with single-filter printing was found in this corpus, so the claim that split-grade printing is sharper — and the claim that it is not — are both dropped. What can be said is what the light path does: both exposures pass through the same negative, the same lens and the same aperture, so nothing about splitting an exposure into two changes the image-forming optics. The one filtration effect on sharpness that is documented has nothing to do with splitting — a filter below the lens is in the image-forming path, and ILFORD state that marks on a filter degrade the image. Two filters below the lens are two opportunities for that, which is a mild argument for the filter drawer rather than for or against the technique.

It does not rescue a negative with no separation to give. If two adjacent tones have the same density on the film, no combination of exposures produces a difference between them on the paper. Filtration redistributes what the negative holds; it cannot add what was never recorded. That is the same limit reading the negative set, and no printing technique on this course moves it.

Local contrast, which is a different quantity

Overall contrast is a property of the whole scale: how much exposure separates the print’s white from its black. Local contrast is the separation between two adjacent tones, and it can be increased in one part of a print while the ends of the scale stay exactly where they are.

That distinction is what makes the difference between a print that measures correctly and a print that looks like something. A negative can fit filter 2 perfectly — highlight and shadow both landing where they should — and still look flat, because the two mid-tones the picture is actually about are a tenth of a stop apart and the paper has rendered them a tenth of a stop apart, faithfully and uselessly.

Three moves change local contrast without moving the ends, and they belong to three different pages.

• A hard-filter burn on a bounded area, which is this page’s contribution and the local-control lab’s practice. It steepens the relationship between tones inside that area, at the cost of darkening it.
• Flashing, treated below, which flattens the highlights alone.
• Development and paper choice, which set the shape of the curve that everything else works on and belong to Part XVIII.

The vocabulary for placing tones on the scale — where a mid-tone should sit, what “one zone” means as an exposure — is Part XIII’s and is used here without being re-derived.

Contrast controls that are not filtration

Filtration is the designed control and the largest one. Five others exist, and each has a cost worth stating plainly, because three of them are commonly used to do jobs they cannot do.

Handle	What it moves	What it costs
Developer choice	Contrast a little, image colour a lot	Changes tone and effective speed with it
Developer dilution	Development time and control; on Foma’s warm-tone papers, colour	ILFORD publish no contrast claim for paper dilution
Development time	Image colour; not contrast	Underdevelopment reads as weak mottled blacks, not as a lighter print
Flashing / pre-exposure	Highlights only, softer	Costs highlight brilliance, and it is one-way
Bleaching the finished print	Density locally, after the fact	Irreversible; Level B chemistry; permanence consequences unestablished

Developer choice is real but modest, and it moves colour more than contrast. Foma sell FOMA GD-L as a “contrasting” phenidone–hydroquinone developer for graphic materials and papers, which is a maker’s own statement that a developer can be a contrast decision. Their note on FOMATONE runs the other way and is the more useful one: developers “giving more expressed images usually reduce contrast and the yield of speed”. ILFORD’s three paper developers are distinguished in their own literature by tone rather than by contrast — MULTIGRADE neutral, PQ UNIVERSAL and BROMOPHEN slightly warm to neutral.

Dilution is a control over the development, not over the contrast. ILFORD offer MULTIGRADE developer at 1+9 and at 1+14 and describe the second as being for “greater development control and economy”; they make no contrast claim for it, and neither does this page. Foma do publish a dilution effect, and it is a colour effect: FOMATOL PW diluted 1+1 with the time doubled, or 1+3 with the time roughly quadrupled, gives warmer image tones on the Fomatone papers. The chemistry is Part XVIII’s.

Development time is not a contrast control on paper, and the maker says so. ILFORD state that development may be extended to six minutes “without any noticeable change in contrast or fog”. A print is developed to completion; a film is pulled at a chosen contrast; the two are not the same operation and the habit does not transfer. What extending development does move is colour — ILFORD recommend approximately double the standard times with MULTIGRADE RC COOLTONE “to obtain the coolest image colour”. So the printer who lengthens development to add contrast has changed the print’s colour and left its contrast alone.

Flashing and pre-exposure

A small uniform exposure, below the paper’s threshold on its own, lifts the whole sheet to the foot of its curve, so that the faintest image exposures land on a responsive part of the scale instead of on the toe. The highlights compress; the shadows, already far up the curve, are unaffected. It is a change in curve shape that filtration cannot make, and it is one-way: it can soften highlights and cannot harden them.

The two words mean the same operation and are used interchangeably by the one maker who documents it. HARMAN’s sheet for Direct Positive Paper heads the technique “pre-flash” and then describes finding the time by “pre-exposing a small sheet in its entirety”. Where the course keeps them apart it is by what receives the light: pre-exposure usually names the uniform exposure given to a film before the camera exposure, flashing the uniform exposure given to paper. On paper the order is the printer’s choice and the sum is the same either way, by the same additivity as a split pair.

=The only manufacturer procedure in the corpus, and what it says about ordinary paper

Searching the corpus for flashing guidance on current enlarging papers returns nothing: ILFORD’s MULTIGRADE sheets, Kodak’s printing publications and Foma’s paper sheets do not mention it. One current product document does, and it is worth reading carefully because it describes the ordinary case in passing.

HARMAN Direct Positive Paper is a fixed-grade paper, and its sheet gives a full pre-flash procedure: make a test strip “in a similar manner to that which could be carried out when using standard photo paper such as ILFORD MULTIGRADE FB”, then select a step, then give the whole sheet that exact exposure. Because the paper is a direct positive, the step to select is the one just showing the slightest shadow detail in the black — and the sheet adds the parenthesis that matters here: “this is the opposite to how ‘pre-flashing’ works with conventional photo paper – where one would select the white which is just showing the slightest indication of highlight detail”.

That parenthesis is the criterion, from a Tier 1 source. On ordinary enlarging paper, the flash exposure is the one at which base white first shows the faintest tone — the largest uniform exposure that still leaves the paper looking white. More than that and the print has grey highlights and no brilliance left in it.

The sheet’s practical notes travel too: an enlarger with a defined aperture and a precise timer is the instrument for finding the time, and a desk lamp will do it in “1 or 2 seconds” at the price of much more trial and error. No figure for a conventional enlarging paper is published anywhere in this corpus, and none is offered here. It is a measurement, and it belongs on the printing map with the paper and developer it was measured for.

Bleaching after the fact

A ferricyanide bleach removes silver from a print that has already been made, which lightens what it touches. Applied locally it is the only handle on this list that works after the paper has been fixed, and the only one that cannot be undone.

Where it belongs in this course. The chemistry and the quantities belong to the formulary, and they are already there: Kodak R-4a is Farmer’s reducer as Kodak Limited printed it in 1949, a ferricyanide stock and a hypo stock mixed one part to four to twenty-seven at the moment of use, because the mixed bath decomposes. Two things about that entry belong here rather than there. It is classified Level B, not Level A like everything else in this part, and the reason is not the intended reaction: the international chemical safety card for potassium ferricyanide states that the substance decomposes on heating to produce toxic gases including hydrogen cyanide, and that it reacts with acids, which generates a toxic hazard. So a bleach tray and an acid stop bath must not share a bench run, a pair of tongs or a waste container. And Kodak’s own directions are written for a negative in a dish; the formulary entry records the bath for film, plate and paper alike, but the local application of it to a finished print is a working practice for which this course has no Tier 1 procedure, and it is named as such. Toning belongs to Part XX and is a different operation — bleaching removes silver, toning converts it.

What the course cannot tell you is the permanence consequence. No conservation or manufacturer source in this corpus establishes what local ferricyanide bleaching does to the long-term stability of a silver gelatin print — whether residual ferrocyanide is left in the paper, at what level, and what it does over decades. That is a real gap and it is stated rather than filled. The practical consequence is a recommendation and it is the course’s: treat bleaching as a decision about a print you have already decided to make, not as a routine contrast control, and reach for the exposure first, because an exposure is repeatable and a bleach is not.

Image tone, filtration, and the record that survives

On a neutral paper, filtration changes contrast and leaves the colour of the image alone. On a warm-tone paper it is not that tidy, and the numbers show why.

Warm-tone papers take the full-stop version of the filter-4 step, and they are the slowest papers on ILFORD’s table. Both warm-tone products drop from P100 to P50 between filter 3 and filter 4 — a full stop, the same step MULTIGRADE IV RC DELUXE and RC COOLTONE take, against an eighth of a stop on FB CLASSIC and the current RC DELUXE. They start from half those papers’ speed as well. So the printer who reaches for a hard grade on a warm-tone paper is making the longest exposure of the evening by a wide margin, and on a paper as slow as FOMATONE, at a small aperture and by contact, that is the direction Foma’s Schwarzschild warning points in. It is a warning about exposures approaching fifteen minutes, which an ordinary enlargement is not — but doubling an already long exposure is how a printer gets there.

On FOMATONE the exposure factor moves inside the soft range as well, by three-quarters of a stop between filters 00 and 3, so there is no part of that paper’s range where a strip transfers without re-anchoring.

And the paper itself is part of the colour. Foma state that the FOMATONE base is “coloured in compliance with the tone of the developed silver”, which is why a warm-tone print looks warm before any toner is near it — the base white is warm, and every tone is seen against it. Meanwhile the developer and the development time are both moving the colour: BROMOPHEN is ILFORD’s own recommendation for the warmest tone on MULTIGRADE Warmtone papers, Foma name FOMATOL PW and its dilutions for the warmest on FOMATONE, and approximately doubled development is ILFORD’s instruction for the coolest colour on COOLTONE. Part XVIII owns the mechanism.

What is not established is that the filter itself moves the colour. No source in this course’s corpus says that changing filtration changes the image tone of a variable-contrast paper, and the published description of the material gives no obvious route for it: ILFORD’s three components differ in green sensitising dye and in nothing else, not in halide or in crystal size. What filtration certainly does is change the exposure, sometimes by a stop, and on a warm-tone paper exposure and development are the two things a printer trades against each other. So the practical warning stands even though the mechanism is unestablished: change the grade, the developer and the development time in one session and you will not know which of them moved the colour. Change one at a time, and write down which.

📓The contrast fields on the printing map, and why a split pair needs six of them

A single-grade print needs one contrast field. A split pair needs six, and every one of them is load-bearing when you come back in a month.

1. Paper, surface and batch. The range and speed tables are per product, and ILFORD’s own note that filters made for Multigrade II and III are “not suitable and should be replaced” is a reminder that even the filters have a generation.
2. Filter set and light source. Which filters, above or below the lens, and which lamp. Foma publish two different tables for one sheet of paper under two filter sets; the grade number alone is not a record.
3. The two black-point anchors, t_hard and t_soft, with the date they were measured. Everything else on the map is a fraction of these.
4. The two times, H and S, in the order they were given, with the order stated.
5. The derived pair, r and q. These are what survive a change of enlargement, and they are what you compare between sessions.
6. The single-grade equivalent, from your calibration table, so that the map can be read by somebody who prints with one filter.

Add the developer, its dilution, its age and the development time in the same block rather than elsewhere, because on a warm-tone paper those four fields are part of the contrast record and the whole of the colour record.

Alternative route

Most of this page is arithmetic. Reading a maker’s speed and range tables, converting a factor to stops, computing r and q from two times and running the conversion the other way are all daylight work with a calculator, and they are the part of the page that changes how you print.

Three things on this page need a darkened space — the two black-point anchors, the calibration prints and the curve-shape experiment — and they need it for the same reason everything else in this part does: the paper must be in the dark from the moment the packet is opened until it is fixed.

Without a permanent darkroom. All three are an evening’s work in any room that can be blacked out, with the wash done afterwards in daylight. Part XVI’s reversible blackout is enough.

Without an enlarger. Everything on this page survives being done by contact at negative size with Part XVI’s frame, and split-grade printing survives it particularly well: the filter goes between the light source and the frame, the two exposures are timed exactly as they would be under an enlarger, and r and q mean the same thing. What shrinks is the local work — a hard-filter burn on a bounded area is clumsy when the whole print is the size of the negative — so the calibration and the arithmetic transfer intact and the local-contrast argument has to be taken on trust until there is an enlarger.

Without any darkened space at all. The honest answer is that the two anchors cannot be measured, and without them r and q have no scale. What can still be done in full is everything that does not need them: read your paper’s published range and speed tables, convert every step in the range to stops, work out what a strip made at one grade would cost you at another, and follow the worked example above on paper with the manufacturer’s own figures. Write the map with the exposure fields blank. A map whose contrast arithmetic is complete and whose two anchor times are honestly empty is a real document; one with estimated anchors in it is worse than a blank one, because a later reader cannot tell which numbers were measured.

Recap

• The grade is chosen twice: once from the negative’s measured range against the maker’s published ISO(R) table, and once from what the picture needs. The second choice may overrule the first, and the cost goes on the map.
• A filter selects components, not contrast. Blue exposes all three components together and stacks their curves; green exposes them in sequence and staggers them. Magenta is the hard direction, yellow the soft.
• Speed matching is a property of a paper and a filter set together. ILFORD’s speed tables are flat from filter 00 to 3 and step between 3½ and 4, by a full stop on the warm-tone papers and about an eighth of a stop on FB CLASSIC and the current RC DELUXE. FOMATONE under ILFORD filters is neither flat nor monotonic and moves 0.77 of a stop inside the supposedly flat range.
• A lengthening factor is the reciprocal of the speed ratio, which Foma’s two tables confirm to the last digit — so any published speed table can be read as a factor table.
• Split-grade printing is two additive exposures, and the latent image on these papers is stable for 24 hours, so a filter change costs nothing. Reciprocity at long exposures is the caveat that does bite, and Foma name it.
• The mechanism predicts no curve-shape difference between a split pair and the single filter delivering the same blue and green quantities. Nobody in the corpus has measured it either way; the test is stated and the page depends on none of it.
• Grade is the hard fraction r, density is the total q. Two measured black points make both computable, both reversible, and both portable between enlargements.
• Split-grade earns its place in local control, where a burn given through one filter changes local contrast as well as density. It does not extend the range, is not established as sharper, and cannot invent separation the negative never held.
• Contrast controls that are not filtration all cost something: developer choice moves colour more than contrast, dilution has no published contrast claim on paper, development time is a colour control and not a contrast one, flashing softens highlights one way only, and bleaching is irreversible Level B chemistry with unestablished permanence consequences.
• The origin of split-grade printing could not be established from any Tier 1 or Tier 2 source, and no attribution is made here.

Check your understanding

Question 1. A test strip is made at filter 2 and works out at 12.0 s. You decide the print needs filter 4½. On ILFORD MULTIGRADE FB WARMTONE, what does the published data tell you to do with the exposure, and what does the same data say on MULTIGRADE FB CLASSIC?

• Nothing on either paper — ILFORD’s filters are speed matched across the whole range
• On FB WARMTONE the ISO speed drops from P100 to P50, a full stop, so roughly 24 s; on FB CLASSIC the speed table drops from P230 to P210, about an eighth of a stop, but that sheet’s own prose asks for 1.5 to 2 times, so the two ILFORD figures disagree and you measure your own
• Double the exposure on both papers, because filters 4 to 5 always need double
• Halve the exposure on both, because a harder grade needs less light

Show the answer and why

Answer: On FB WARMTONE the ISO speed drops from P100 to P50, a full stop, so roughly 24 s; on FB CLASSIC the speed table drops from P230 to P210, about an eighth of a stop, but that sheet’s own prose asks for 1.5 to 2 times, so the two ILFORD figures disagree and you measure your own

Speed matching runs from 00 to 3½ only, and the size of the step above it is a property of the product. The warm-tone papers step by exactly the factor of two the contrast-control sheet states. FB CLASSIC does not: its speed table implies 0.13 of a stop and its prose asks for 0.6 to 1 stop. They are not measuring the same thing — an ISO speed is defined at one density point, a practical filter factor is what keeps a whole print looking the same — and the resolution is not to choose one but to measure the factor for your own box and write it where you will find it.

Question 2. You print a sheet with 4.8 s through filter 5 and 30.0 s through filter 00. Your measured black points are 12.0 s at filter 5 and 60.0 s at filter 00. What are the hard fraction and the total dose, and what would you change to make the same grade half a stop lighter?

• r = 0.44 and q = 0.90; multiply both times by 0.707, giving 3.4 s and 21.2 s
• r = 0.44 and q = 0.90; reduce only the hard time, to 2.4 s
• r = 0.16 and q = 34.8; subtract 2.4 s from each time
• r = 0.50 and q = 0.90; halve the soft time, to 15.0 s

Show the answer and why

Answer: r = 0.44 and q = 0.90; multiply both times by 0.707, giving 3.4 s and 21.2 s

h = 4.8 ÷ 12.0 = 0.40 and s = 30.0 ÷ 60.0 = 0.50, so q = 0.90 and r = 0.40 ÷ 0.90 = 0.44. Half a stop lighter is a factor of 2 to the power −0.5, which is 0.707 applied to the total; because it is applied to both times equally, r is untouched and the grade holds. Reducing only one of the two times changes the blue-to-green ratio, which changes the grade as well as the density — the mistake the two-number bookkeeping exists to prevent. Note also that r is not the ratio of the two raw times: 4.8 against 30.0 looks like a mostly soft print, and it is, but only after each time has been divided by its own filter’s black point does the number mean anything.

Question 3. Why does a burn given through the hard filter alone do something a burn at the print’s overall grade cannot?

• Because the hard filter transmits more light, so the burn is quicker
• Because the extra exposure goes almost entirely to the components that answer to blue, so it steepens the separation between tones inside the burned area as well as darkening it
• Because magenta filters make the paper more sensitive
• Because a second exposure always produces a different latent image from a single one

Show the answer and why

Answer: Because the extra exposure goes almost entirely to the components that answer to blue, so it steepens the separation between tones inside the burned area as well as darkening it

All three emulsion components share one blue speed, so a blue-rich exposure adds the same amount to each and their curves stay stacked — which is the high-contrast condition. Adding that exposure to a bounded area changes the local relationship between tones there, not just their average density, which is the definition of a local contrast move. A burn at the print’s overall grade adds blue and green in the print’s existing proportion and so only darkens. The hard filter transmits less, not more, which is why grade 5 costs exposure on most papers.

Question 4. A printer claims that split-grade printing gives a curve shape no single filter can produce. What does this course say about that claim, and why?

• It is established: manufacturers publish measured curves for split exposures
• It is false: the course has measured it and found no difference
• It is unestablished, and the published mechanism argues against it — since the three components share one blue speed and differ only in green, a split pair delivers the same blue and green quantities as some single mixed exposure, so no shape difference is predicted; no source in the corpus has measured it either way, and the test is a step wedge exposed both ways
• It is irrelevant, because contrast cannot be measured on paper

Show the answer and why

Answer: It is unestablished, and the published mechanism argues against it — since the three components share one blue speed and differ only in green, a split pair delivers the same blue and green quantities as some single mixed exposure, so no shape difference is predicted; no source in the corpus has measured it either way, and the test is a step wedge exposed both ways

This is a case where the absence of evidence and the presence of a mechanism point the same way. ILFORD’s own account reduces the light to two quantities the paper can distinguish, and two exposures adding to a total is arithmetic, so the paper should not be able to tell a split pair from the equivalent single mixture. What that argument assumes is stated: the three-component description is the maker’s simplification, additivity across a filter change is inferred from a 24-hour latent-image statement rather than measured, and reciprocity is known to fail at long exposures on at least one paper. The right response to an unmeasured claim is a measurement, not a stronger assertion in either direction.

Question 5. Your prints are coming out flat. You extend development from 90 seconds to 3 minutes and nothing much changes except that the image looks slightly colder. What has happened?

• The developer is exhausted and needs replacing
• Development time is not a contrast control on paper — ILFORD state that development may be extended to six minutes without any noticeable change in contrast or fog — but it does move image colour, which is why ILFORD ask for roughly double the time on COOLTONE to get the coolest image colour
• The paper has been fogged by the safelight
• The exposure was too short, and longer development compensates for it

Show the answer and why

Answer: Development time is not a contrast control on paper — ILFORD state that development may be extended to six minutes without any noticeable change in contrast or fog — but it does move image colour, which is why ILFORD ask for roughly double the time on COOLTONE to get the coolest image colour

A print is developed to completion, unlike a film, which is pulled at a chosen contrast; the plateau is what the six-minute statement describes. So extending development past the recommended time changes almost nothing about contrast and does change the colour of the deposited silver. Genuine underdevelopment looks different again — weak, mottled blacks rather than an evenly lighter print. The handle for flatness is filtration, and after that the negative.

Check my answersClear