This case involves an action for contribution under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA). The Boeing Company and Cascade Corporation contaminated porous soil, called an aquifer. The contamination overlapped. Both companies spent a great deal of money investigating and beginning the cleaning. The issue in this case is how their respective expenses ought to be allocated in a contribution action.

FACTS

Boeing sued Cascade for contribution to offset Boeing’s higher expenses so far, and for a declaratory judgment allocating future expenses. The case was tried to the court without a jury. Boeing won, though not all it sought. Both sides appeal.

The district court determined that the fairest way to allocate costs between the companies was by the quantity of toxic chemicals each company put into the ground. Cascade was judged liable for seventy percent of the costs, Boeing for thirty percent. Cascade argues that the 70:30 allocation is erroneous and that the district court lacked jurisdiction to enter a declaratory judgment imposing the 70:30 ratio on future expenses. Boeing argues that Cascade was allowed to apply too much of what it had spent to its seventy percent share, and was erroneously given credit against its share for a settlement Boeing collected from a third party. We affirm in part and remand in part the district court’s judgment.

Boeing owns land about 200 feet northwest of Cascade’s land near Portland, Oregon. Because the ground slopes down to the north, liquid in the ground flows from Cascade’s property towards Boeing’s. Boeing manufactures airplane parts on its 151 acres and Cascade manufactures lift truck attachments on its 6 acres. Both companies and their predecessors have used chlorine-based solvents to clean machine parts and for vapor degreasers. Some of the solvents spilled or were dumped and contaminated the soil at each site. Nobody claims that there were any laws prohibiting them from disposing of the solvents in such a way at the time. Public authorities eventually grew concerned that contaminated groundwater flowed north towards the Columbia River, and towards water wells operated by the City of Portland as a supplemental source of drinking water. Neither Boeing nor Cascade has been accused of any legal wrongdoing and the evidence indicates that both companies have been diligently attempting to remedy the problem since learning of it.

In 1985, Boeing drilled monitoring wells on its property for another project and discovered the contaminated aquifer. An aquifer is a relatively porous soil layer, such as sandstone, that holds water, something like a sponge. In 1986, Boeing signed a consent order with the Environmental Protection Agency requiring Boeing to investigate the extent of the contamination and to identify its potential sources. Boeing installed monitoring wells, sampled groundwater and soil gas, and conducted pump and tracer tests to determine the sources and extent of the contamination. Sixteen of the wells, in Boeing’s southeast corner (the part of its property closest to Cascade’s), yielded samples suggesting that contaminated groundwater was flowing from Cascade’s property into Boeing’s.

Cascade learned of contamination in its industrial supply well a little later than Boeing, in 1986. For the next two years, Cascade tested the well and sent the results to the Oregon State Department of Environmental Quality. In 1987, Cascade tested the waste in the underground tanks it used to store waste oil and coolants, and discovered chlorine-based chemicals in the waste. Cascade removed the contaminated soil from the site and, in 1988, entered into a consent order with the Department of Environmental Quality to evaluate the extent and source of contamination at its site. Cascade, like Boeing, investigated the potential sources of the contamination, and has been trying to eliminate it.

In 1989, Boeing and Cascade began working together on investigation and cleanup under the supervision of the Oregon Department of Environmental Quality. In 1998, Boeing and Cascade entered into a joint consent order that directed the parties to control the contaminant plume and to learn more about the area so as to protect Portland’s water supply. Their cooperation has continued since.

There are three aquifers separated by two confining units or aquitards. An aqui-tard is a relatively impermeable layer that retards the flow of water. Thus, the structure is something like three wet sponges separated by dishes. The top sponge is called the Troutdale gravel aquifer. Beneath it is an aquitard, then the second sponge, called the Troutdale sandstone aquifer. The sandstone layer has an upper more uniform part, and a lower, more varied part. A second aquitard lies below it, then a third aquifer. The second aquifer down, the sandstone layer, holds the Boeing.and Cascade contamination at issue.

Because the aquifers and aquitards are natural dirt and rock, not sponges and dishes, they are irregular. Each layer contains “lenses,” or points within it that may hold water or retard its flow differently from the surrounding oil and rock. Some water flows along an aquitard, then seeps around a lens and through breaks and discontinuities into the aquifer below. In some areas, erosion and geologic stresses have removed the top aquifer and aquitard, leaving the sandstone aquifer immediately below the surface. Over the northeast portion of the Boeing land the top aquitard disappears, so groundwater flowing north toward the river in the gravel aquifer on top seeps without hindrance into the sandstone aquifer below it. The slope of the underground formations makes the water flow south instead of north for a stretch, then breaks it into two separate flows north toward the river.

There is a north-south separation in the sandstone aquifer on Boeing’s property. The parties agree that groundwater west of it has been contaminated solely by Boeing, while groundwater east of it was contaminated by Cascade and possibly by Boeing as well. One of Boeing’s experts, Dr. Henry Landau, testified that pumping from wells operated by the city may have pulled some of the contaminants across the separation. When a well is pumping, it pulls water toward it. The City of Portland’s wells draw groundwater toward them from more than two miles away. A local water district, the Rockwood Wells District, also draws from wells in the area and pulls water. Boeing started to pump and treat the groundwater in 1989, pulling some of the contaminated groundwater from the gravel aquifer so that it did not flow into the sandstone aquifer.

Both the gravel and sandstone aquifers are contaminated. Boeing and Cascade are separately cleaning up the gravel aquifers on their respective property and those costs are not at issue in this case. But contamination from the gravel aquifer has seeped past the top aquitard into the sandstone, as explained above. The dirty water plume extends beyond Boeing’s and Cascade’s properties and may have seeped past the second aquitard into the third aquifer down, the sand and gravel aquifer. The flow of contaminants could eventually reach the City of Portland wells if nothing were done about it.

The contamination of the sandstone aquifer is the subject of this case. The contaminant plume is generally north of the Cascade property, and extends both northeast and northwest of Cascade. The northwest portion of the plume extends under part of Boeing’s property. The eastern portion of the plume contains groundwater that flowed north from Cascade, and is solely attributable to Cascade. The westernmost portion of the plume is solely attributable to Boeing and its predecessors. A portion of the plume in between contains contaminated groundwater from both Boeing and Cascade.

The .contaminants in the groundwater are chlorinated solvents known as volatile organic compounds. The major contaminants in the Sandstone Aquifer are trichlo-roethene, 1,2-dichloroethene, and perehlo-roethene or tetrachloroethane. Freon, 1,1,1-triehloroethane, and 1,1-diehloroeth-ene are also present in lesser concentrations. These compounds, the findings say, share common properties such as high toxicity, low solubility, and high mobility in the environment.

The contaminants came from solvents used by Boeing and its predecessors and Cascade. Boeing and its predecessors used trichloroethene as their primary de-greaser solvent until 1980, and then switched to 1,1,1-triehloroethane. They stored waste solvents in storage tanks and dumped them into the ground or stored them in barrels underground. Based on evidence available at trial, the original contaminant mass in the gravel aquifer at Boeing was about 4500 pounds. As of the time of trial, 2560 pounds had been extracted, leaving 2046 pounds.

Cascade’s vapor degreaser used trichlo-roethene from 1961 until 1975 when it switched to non-chlorinated solvents. Cascade also stored waste solvents in storage tanks and may have dumped some of the waste into the ground.

ANALYSIS

A. Causation.

Cascade argues that it should not have been held liable for a share of Boeing’s costs because Boeing would have incurred them even if Cascade had not contaminated Boeing’s soil. Since the EPA ordered Boeing to investigate its own contamination, Cascade reasons that Boeing would have incurred its investigation costs even without Cascade’s releases, and therefore Cascade was not the cause of the response costs incurred by Boeing. CERCLA conditions liability on causation, and in Cascade’s view, its contamination cannot be a cause but for which Boeing would not have incurred costs, when Boeing would have had to incur the same costs without it.

We review the district court’s interpretation of the statute de novo CERC-LA provides that a party that releases a hazardous substance is liable for another’s response costs, but only if its release caused the other party to incur those response costs:

The statute says that a party that releases hazardous substances is hable if the release “causes the incurrence of response costs.”

The essence of Cascade’s argument is: (1) we can be liable only if our release of contaminants caused the response costs; (2) Boeing would have had the response costs anyway, i.e., because of its own release of contaminants it cannot be said that the costs would not have been incurred “but for” our release of contaminants; so (8) we cannot be liable, since our release did not cause the response costs.

Causation requirements of course permeate many areas of the law. The statute requires a releaser to compensate a party that bore response costs only if the release “causes the incurrence of response costs.” The issue Cascade raises is whether causation, which the statute expressly requires, is necessarily “but for” causation. “But for” causation is a short way of saying “[t]he defendant’s conduct is a cause of the event if the event would not have occurred but for that conduct.” It is sometimes stated as “sine qua non ” causation, i.e., “without which not,” and that may be clearer because it does not require that the reader remember the rest of the “but for” sentence. The entirely undescriptive term “causation in fact” has sometimes been used to mean the same thing.

The courts and scholars have long recognized that there is a limited class of cases in which conduct ought to be treated as a cause of damages even though the conduct was not a sine qua non. For example, Harper and James cite the case of two motorcyclists who pass a horse, one close to each side. The spooked horse gallops out of control and causes injury. Neither motorcyclist’s conduct was a sine qua non of the injury, because the horse would have been spooked by the other motorcyclist. The rule that has evolved is that, at least where both causes involve comparable blameworthiness, both actors are liable, even though the conduct of either one was not a sine qua non of the injury because of the conduct of the other. There is no reason why a polluter should be insulated from responsibility in a case where a traditional tortfeasor would not be.

Cascade would have us read the statutory language, “causes the incurrence of response costs,” to mean that liability cannot be imposed without sine qua non causation. That raises the question whether the word “causes” means “causes as a sine qua non.” Ordinarily it does, but we have already explained that in common law tort actions, it sometimes does not. Hart and Honoré, in their classic discussion of causation in the law, describe a class of cases “when causally relevant factors are not conditions sine qua non.”

Hart and Honoré criticize Prosser’s question begging formulation, that “a cause in fact” is a “substantial” factor, and “ordinarily conduct will be a substantial factor if the result would not have happened without it.... Little, however seems to be gained by describing, even to a jury, such cases [in which neither cause is a sine qua non ] in terms of the admittedly indefinable idea of a ‘substantial factor’.” Hart and Honoré ultimately conclude that “when each factor is sufficient, with other normal conditions, to bring about the harm as and when it occurs, each is properly described as a cause of the harm,” and thus both actors are liable.

Philosophers subsequent to Hart and Honoré have developed the analysis of causes that are not sine qua nons under the rubric of “causal overdetermination.” Two classic examples they use are these:

The philosophers’ phrase entered the law in Justice Brennan’s plurality opinion in Price Waterhouse v. Hopkins, where he refers to “events that are causally overdetermined.” The term is useful, because it distinguishes the small number of cases in which a legal cause need not be a sine qua non from the general run of cases where a cause has to have “made a difference to the outcome” and “any standard less than but-for ... simply represents a decision to impose liability without causation.”

In the special circumstance of causal overdetermination, conduct can be a cause of a result even though it is not a sine qua non. Take the philosophers’ example above of the kitchen with a light switch at each end. When two people simultaneously flip both switches on, the light goes on. Neither person’s conduct is a sine qua non, because the light would have gone on anyway. Neither individual's conduct made a difference to the outcome. Cascade’s analysis would compel the conclusion that neither person caused the light to go on.

Cascade’s argument that liability can only attach to conduct that is a sine qua non of the harm, even for causally overdetermined harm, cannot be right, as the kitchen light hypothetical case shows. The problem with Cascade’s argument is that where the result is overdetermined, each person’s argument is as strong as the other’s identical argument. If we accept one person’s argument that he did not cause the light to go on, then we have to accept the identical and equally valid argument of the other person that he did not cause the light to go on. Each accurately points out that his switching the light on was not a sine qua non of its going on. It is true that the light would have gone on anyway because of the other person’s conduct. If conduct had to be a sine qua non even for this overdetermined result, then neither person’s conduct caused the light to go on. But the light went on. And it did so by human agency, not spontaneously. So the conclusion that Cascade’s argument compels, that no one caused the light to go on, is false. Because the correct answer has to be the same for the two individuals, by eliminating the false answer we have left only one possible answer which must be true: Each of the two persons caused the light to go on.

This is not just a matter of abstraction. Cascade’s argument, if accepted, would generate senseless practical consequences. Were we to accept Cascade’s argument that sine qua non causation is required even in cases of causal overdetermination, CERCLA would encourage parties to avoid investigating and cleaning up contamination to which others as well as themselves contributed. A party that had discharged into a mixed plume could wait for another discharger to incur the costs of investigation, and have a fair chance of leaving the other polluter stuck with the entire bill. Because the pollutant, rocks, and dirt do not respect property lines, the expense of cleanup may be only slightly responsive to quantity, causal overdetermi-nation of response costs may be less rare in CERCLA cases than in other kinds of cases. A requirement of sine qua non causation in such cases would give irresponsible defendants an absolute defense. One of the goals of CERCLA was to “affix the ultimate cost of cleaning up these disposal sites to the parties responsible for the contamination.” To leave one party shouldering the entire cost of investigation and remediation while another rides for free frustrates this goal, rather than ensuring that those who caused the contamination pay their fair share of the costs associated with clean-up. Thus, the practical consequences lead the same way as philosophical analysis in construing the statutory requirement of causation.

We therefore conclude that in the special case of causal overdetermination, i.e., where either polluter’s conduct would have caused the same response cost to be incurred in the same amount, and the conduct was of substantially equal blameworthiness, the proper construction of the causation requirement in the statute is that both polluters should be treated as having caused the response cost. In this special case, pollution is properly treated under the statute as having “caused” the response cost even though it wa's not a sine qua non of the response cost.

Other circuits have likewise construed CERCLA to mean that once a party is liable, it is required to share the costs of response regardless of whether it is the sole cause of those costs. In Control Data Corp. v. S.C.S.C. Corp., the Eighth Circuit resolved this issue as we have. The Eighth Circuit noted CERCLA’s dual goals of “quick response and to place the cost of that response on those responsible for the hazardous condition,” because a party discovering contamination would not want to investigate if it could not easily recover those costs. Control Data cited as support the Supreme Court’s decision in Key Tronic Corp. v. United States, which held that costs attributable to searching for other responsible solvent polluters are recoverable, reasoning that such an investigation promotes cleanup. Cascade seeks to distinguish Control Data on its facts. But, despite the distinctions, Control Data’s reasoning applies to the facts before us. Likewise, in Browning-Ferris Industries, the Seventh Circuit resolved this issue as we do. It held that so long as the defendant’s pollution “was serious enough ... to require that the site be cleaned up,” the defendant is still a sufficient cause of the clean up and therefore “it is not inequitable to make it contribute substantially to the cost.”

B. Allocation of Costs.

1. Accounting.

Cascade argues that Boeing did not properly separate its expenses for the gravel aquifer from those for the sandstone aquifer. Because Boeing’s claim only related to the sandstone layer, Cascade argues that the district court erroneously burdened it with 70% of about $2 million incorrectly attributed to the sandstone aquifer. Boeing did not keep separate accounts for the gravel and sandstone aquifers when it spent the money. Instead, its expert witnesses made an allocation after the fact, which the judge evaluated in light of all the evidence, including Cascade’s expert’s views.

To recover damages, CERCLA requires a plaintiff to prove that costs are “necessary response costs consistent with the national contingency plan.” The national contingency plan requires “accurate accounting of ... private party costs incurred for response actions.” But “accurate accounting” need not mean contemporaneous separation of expenses. No case has held that it does. Neither party knew that the other was involved at the time the contamination was discovered, so there was no reason to segregate costs. Post hoc accounting lends itself to inappropriate shifting of expenses from noncompensable to compensable categories, but so would contemporaneous accounting, for example, where one company knew one category might be compensable but not the other. The contemporaneity of the accounting, and the interests a party might be trying to advance when it did the accounting go only to the credibility of the accounting.

Boeing presented extensive testimony about its investigation costs. It provided exhibits illustrating its total costs, both internal and external. One of its witnesses, the accounting consultant, described his function as “accounting] for costs and to assure they are reasonably accurate and necessary.” He explained Boeing’s accounting procedures and referred to exhibits summarizing five boxes of invoices. Boeing’s spreadsheets broke out the separate costs for the gravel aquifer and the sandstone aquifer based upon the allocation decisions of its accounting consultant.

The accounting consultant testified that he segregated any costs that were exclusively related to the gravel aquifer or the sandstone aquifer. Then he allocated indirect costs using the average, 37.2%, of the percentage of wells drilled into the sandstone aquifer (34.3%) and the percentage of samples that were collected from the sandstone aquifer (40.1%). Thus, 37.2% of Boeing’s indirect costs were attributable to the sandstone aquifer, or $1,984,825. The direct costs, $4,231,634, added to the indirect costs, $1,984,825, plus pre-judgment interest of $1,043,262, result in a total of $7,259,721 attributable to the sandstone aquifer. Though other approaches might be as well or better justified, such as perhaps allocating each of the 11,000 receipts, the district judge could reasonably give credence to the method that Boeing used. We review the district court’s finding that the remediation of the sandstone aquifer cost Boeing $7.2 million of the $18.4 million it spent for clear error. There was none.

2. Mass.

Cascade argues that the plumes of contaminated groundwater were largely separate, so the district court should not have aggregated the remediation costs and divided the total. Instead, the district court should have left each party with the expenses it bore for the work on its own land. Even if aggregation were proper, Cascade argues that the court erred by dividing the costs according to the volume of each party’s contaminants, 70:30.

CERCLA provides that “the court may allocate response costs among liable parties using such equitable factors as the court determines are appropriate.” This language gives district courts discretion to decide what factors ought to be considered, as well as the duty to allocate costs according to those factors. We reverse only for an abuse of the discretion to select factors, or for clear error in the allocation according to those factors.

Some courts use what are called the “Gore factors,” named after a failed attempt to amend CERCLA. Congress rejected the amendment that would have listed the Gore factors as the basis for allocating liability. The trial court is therefore not limited to the Gore factors. The statutory language gives the trial court the discretion to consider “such equitable factors as it finds appropriate” so we review for abuse of discretion in determining appropriateness, not conformity to some predetermined list or rule. The district court considered the Gore factors, which include volume, but concluded, as is typical with multi-factor tests, that “[m]ost of the Gore factors, unfortunately, fail to assist in this case.” Both parties were generators of the contaminants, neither party’s contaminants were any more toxic than the other’s, and neither party was more careless than the other in fight of the practices characteristic of the time. Moreover, given the practices of the time, it is impossible to know precisely how much waste was dumped, as record keeping was not mandated the way it is today. Once the contamination was discovered and Boeing and Cascade were finked to it, both parties were very responsive in taking steps to begin clean up.

Cascade’s argument that each party should be left with the expense it bore for the cleanup it had done on its own land has no persuasive force. Water flows downhill. Thus, most of the dirty water from Cascade’s uphill site would naturally no longer be there when the cleanup was done, and much of the dirty water in Boeing’s sandstone aquifer would naturally come from its uphill neighbor, Cascade. The district court so found.

Boeing presented extensive expert testimony on the relative mass of each party’s contribution to the plume. An engineer testified for Boeing that he calculated Cascade’s contribution to the total mass of contaminants as eighty percent. To allocate the mass between Boeing and Cascade, he delineated two plumes on either side of the north-south trough: one for which Cascade was responsible, and the other originating at the Boeing site. Thus, he provided the court with evidence from which it could draw a conclusion regarding divisibility and share of responsibility.

Cascade’s expert testified that Boeing’s expert understated Boeing’s contribution to the contaminant mass. She testified that Boeing’s contaminants had flowed into what Boeing’s expert characterized as Cascade’s plume. The district judge accepted her testimony on this point and adjusted the allocation of contaminant mass to reflect Boeing’s larger share: a 70:30 allocation instead of the 80:20 ratio that Boeing’s expert had suggested.

Cascade argues that allocating responsibility simply by volume is inappropriate. This argument, as a general proposition, has persuasive force. Suppose one party disposed of a pound of highly toxic contaminants dissolved in water that flowed over a great expanse, while another dumped a well sealed cannister of 500 pounds of the same contaminant; it might be so much cheaper to haul away the cannister than all the contaminated dirt from the dissolved contaminant that imposing 500 times the expense on the cannister dumper would be inappropriate.

In this case, though, the district court was within its discretion in using volume as the primary or exclusive basis for allocation. Cascade and Boeing each argue for numerous other factors that would have benefitted each. A district judge must use discretion to determine which factors are appropriate in the particular case. We cannot say that this district judge abused his discretion by using volume as the basis for allocation in this case.

The Fifth Circuit has also held that “volume may be a reasonable means of apportioning liability.” “Essentially, the question of whether there is a reasonable basis for apportionment depends on whether there is sufficient evidence from which the court can determine the amount of harm caused by each defendant.” Like the claimant in the Fifth Circuit case, Boeing presented “sufficient evidence from which a reasonable and rational approximation of each defendant’s individual contribution to the contamination can be made.”

Cascade argues that even if allocation by volume was appropriate, the district judge erred in his allocation of seventy percent liability to Cascade and thirty percent to Boeing. We review the district court’s allocation as a finding of fact for clear error. Though the testimony presented might have led the trial court to a different conclusion, the 70:30 allocation was among the reasonable conclusions supported by the evidence. “Where there are two permissible views of the evidence, the fact finder’s choice between them cannot be clearly erroneous.”

3. The Settlement.

Electronic Specialty Company, among others, formerly operated the plant that Boeing now operates. Boeing claimed that it was responsible for much of the contamination. Electronic Specialty and its parent company, International Control Corpo ration, settled with Boeing for $7.5 million. Boeing may receive $6 million more from International Control’s insurers, under a term of the settlement not affecting the issue before us. The settlement was for all contamination, both the gravel aquifer, for which no claim is made against Cascade, and for the sandstone aquifer which is at issue.

The district court found that of the $18 million Boeing had incurred in response costs, approximately one third was for response costs for the sandstone aquifer at issue between Boeing and Cascade. Thus, about $6 million was spent by Boeing on the sandstone aquifer, and about $12 million on response costs for which it made no claim against Cascade. The district court then rejected Boeing’s argument that the $7.5 million it had collected from International Control and Electronic Specialty should not be considered in determining what Cascade owed. The district court opinion says “I find that the prohibition of 42 U.S.C. § 9614 against double recovery requires that the settlement funds be factored into allocation of response costs.”

The district court then allocated the response costs as follow:

The statute says that if a party makes an approved settlement with the federal or a state government, that settlement “reduces the potential liability of the others by the amount of the settlement.” This is of no help to Cascade, though, because Electronic Specialty and International Control settled with Boeing, not with the federal or a state government. The preceding subsection of the statute does speak to this case: “In resolving contribution claims, the court may allocate response costs among liable parties using such equitable factors as the court determines are appropriate.” The district court correctly notes that one equitable factor is preventing someone from recovering for the same harm twice. The statute the district court cited says that a party that receives compensation under CERCLA is “precluded from recovering compensation for the same removal costs or damages or claims” pursuant to other state or federal law. But the numbers in the district court’s calculation do not work consistently with this equitable factor.

The numbers are easier to understand if put in columns like a spreadsheet than spread out in essay form, so here is how we lay them out, consistent with the district court’s unchallenged fact findings:

The question is what to do with the $2,500,000. Electronic Specialty and International Control reimbursed Boeing for that much of the $6,216,459 Boeing spent on the sandstone aquifer. There are no factual issues raised by the parties, just an argument about whether the way the judge treated the $2,500,00 was “double counting.” Here is what the district court did:

Thus Boeing’s judgment against Cascade was for $1,340,437 plus interest.

Laying the numbers out reveals an error. The court determined that Boeing and its predecessors were responsible for only 30% of the pollution in the sandstone aquifer. The total expenditures on the sandstone aquifer were $10,420,074, which is Boeing’s $6,216,459 plus Cascade’s $4,203,615. Although Boeing was reimbursed by its predecessors to the extent of $2.5 million of this $6.2 million expenditure on the sandstone aquifer, so it did not wind up out of pocket $6.2 million, Boeing and its predecessors did wind up out of pocket $6.2 million, and Cascade $4.2 million, for remediation on the sandstone aquifer.

But these figures do not match the district court’s findings that the pollution in the sandstone aquifer came 30% from Boeing and its predecessors, 70% from Cascade. Applying those percentages to the total expenditure of $10,420,074, Boeing and its predecessors should pay $3,126,022, and Cascade should pay $7,294,052. Because Boeing and its predecessors have spent $6,216,459, they have paid $3,090,437 more than their 30% share.

The district court erred in deducting the settlement from Boeing’s expenditure before applying the 70-30 ratio. Instead of eliminating double reimbursement for the same expense to Boeing, which the district court quite properly intended to do, the computation error had the effect of reducing the total expenditure by the $2.5 million Boeing’s predecessors contributed before applying the 70-30 ratio. Thus Boeing and its predecessors wound up stuck with $6.2 million, less the $1.3 million judgment against Cascade, for $4,876,022 of the $10.4 million expense, about 47% of an expense that the district court found was 30% attributable to them. There is no double reimbursement to Boeing for its cleanup expense if the 70-30 ratio is applied to the sum of its and its predecessors’ expenditures plus Cascade’s. Rather, there was a failure to count once the $2.5 million, because the district court did not split that portion of the response costs 70:30.

Because we have rejected the challenge to the 70-30 allocation of responsibility, and there is no challenge (except for the accounting dispute dealt with above) to the 1/3 — 2/3 split between Boeing’s and its predecessors’ expenditures on the sandstone aquifer and the aquifer for which Cascade has no responsibility, the judgment ought to be based on those ratios and the figures above. That implies that Boeing’s judgment against Cascade should be for the difference between the $6,216,459 it and its predecessors spent, and the $3,126,022 they should have spent under the district court’s findings, which is $3,090,437 plus interest. We remand so that the judgment can be corrected to this extent.

4. Internal costs.

[12] Cascade argues that the district court erred by giving Boeing credit for its internal costs as well as what it paid to third parties for investigation and remediation, but giving Cascade credit only for what it paid to third parties. Cascade made no explicit request for recovery of internal costs-not before trial, not after trial, not in any post-trial motion. There was evidence, two pages of testimony by Cascade’s treasurer, that Cascade spent about $1 million internally, based on which the district judge could have credited Cascade. But Cascade did not ask the court to credit it with this amount, or offer an exhibit adding in this amount. The judge evidently used Cascade’s own exhibit showing its expenses as a basis for its findings of fact. Cascade did not make a Rule 59(e) motion or in any way ask the district court to correct the error, if it was an error, by crediting the extra $1 million buried in two pages of its treasurer’s testimony. Because this issue of crediting the internal costs was not preserved by raising it adequately in the trial court, it is not available for appeal.

C. Declaratory Judgment Allocating Future Costs.

The litigation was concluded before the cleanup was concluded. Boeing won a declaratory judgment that the remaining cleanup expenses should also be split 70:30. Cascade argues that the district court lacked authority to issue a declaratory judgment allocating expenses not yet borne, and even if it had authority, the amounts were too speculative to support the judgment. To the extent that this inquiry requires that we construe CERC-LA, our review is de novo.*

We have upheld declaratory judgments in CERCLA cases, but Cascade would distinguish those cases because they concern suits for costs under 42 USC § 9607 , rather than for contribution under 42 U.S.C. § 9613 . They do. Cascade argues that an express statutory provision for declaratory judgments in cost recovery actions in a part of the statute addressing limitations periods implies an absence of such authority in contribution actions, for which there is no parallel language in the statute.

The statute is silent on whether declaratory judgments are authorized in contribution actions. It does not prohibit them. It is hard to see why it would. CERCLA was intended to encourage quick response and to place the costs on those responsible. Declaratory relief serves these purposes because all parties, like those in this case, will know their share of costs before they are incurred. The more liability can be limited and quantified, the more practical it is for a party to budget and borrow to finance it. Environmental litigation is tremendously complex, lengthy, and expensive. The costs and time involved in relitigating issues as complex as these where new costs are incurred would be massive and wasteful. Declaratory relief allocating future costs is therefore consistent with the broader purposes of CERC-LA.

Nor do we see much force to Cascade’s argument based on syntax. If the costs recovery provision said that the court “may” grant declaratory judgments, there might be an implication that it may not in another section without the “may grant” language. But saying, as the statute does, that the court “shall” grant declaratory judgments in cost recovery cases does not imply that it may not in cases where it is silent. Saying that a court “shall” do A, may as reasonably imply that it “may” do B, about which it is silent.

Cascade argues that the facts of this case do not render it appropriate for declaratory relief because the costs of remediation are not sufficiently stable to permit declaratory relief. We review for an abuse of discretion. Whether declaratory relief is appropriate depends upon

That is plainly so in this case. The pollution has been carefully studied, the parties litigated a genuine controversy about millions of dollars they had already spent, and the facts bringing about their relative responsibility have already occurred.

CONCLUSION

We affirm the judgment in response to Cascade’s appeal. On Boeing’s appeal, we remand, so that the amount of the judgment can be increased as explained above in section 3.

AFFIRMED in part and REMANDED in part. Costs in favor of Boeing.

. See 42 U.S.C. § 9607 (a) (1994) (describing liable party as, in pertinent part, an "owner or operator of a facility ... from which there is a release ... which causes the incurrence of response costs.” (emphasis added)). . See Pinal Creek Group v. Newmont Mining Corp., 118 F.3d 1298, 1300 (9th Cir.1997). . 42 U.S.C. § 9607 (a)(4) (1994) (emphasis added). . 42 U.S.C. § 9607 (a)(4); see also 3550 Stevens Creek Assoc. v. Barclays Bank, 915 F.2d 1355, 1358 (9th Cir.1990) (requiring a plaintiff to demonstrate that release or threatened release caused plaintiff to incur response costs). . W. Page Keeton, et al., Prosser & Keeton On the Law Of Torts § 41, at 266 (5th ed.1984). . See H.L.A. Harl & Tony Honoré, Causation in the Law, 128 (2d ed.1985). . See, e.g., Keeton, supra note 6, § 41, at 263. . 4 Fowler V. Harper et al., The Law of Torts § 20.3 (2d ed.1986) (citing Corey v. Havener, 182 Mass. 250 , 65 N.E. 69 (1902)). . See id. . Hart & Honoré, supra note 7, at 122. . Id. at 128-29. . Id. at 124. . See id. at 228-29. . See Louis E. Loeb, Causal Theories and Causal Overdetermination, 71 J. of Phil. 526 passim (1974); see also Martin Bunzl, Causal Overdetermination, 76 J. of Phil. 134 passim (1979); Louis E. Loeb, On a Heady Attempt to Befriend Causal Theories of Knowledge, 29 Phil. Studies 331 passim (1976); Yael Tamir, Who Done It? Moral Responsibility for Collective Action (visited February 29, 2000) http://www.stthom.edu/cbes/zohar.htm (discussing causal overdetermination of results in analysis of voting by members of corporate boards of directors). . Loeb, Causal Overdetermination, supra note 15, at 526. . Price Waterhouse v. Hopkins, 490 U.S. 228 , 109 S.Ct. 1775 , 104 L.Ed.2d 268 (1989). . Id. at 241 , 109 S.Ct. 1775 (emphasis added). . Id. at 282 , 109 S.Ct. 1775 (Kennedy, J., dissenting). . Kaiser Aluminum & Chem. Corp. v. Catellus Dev. Corp., 976 F.2d 1338, 1340 (9th Cir. 1992). . Control Data Corp. v. S.C.S.C. Corp., 53 F.3d 930 (8th Cir. 1995). . See id. at 936 . . Key Tronic Corp., v. United States, 511 U.S. 809 , 114 S.Ct. 1960 , 128 L.Ed.2d 797 (1994). . See Control Data, 53 F.3d at 937 (citing Key Tronic, 511 U.S. at 810 , 114 S.Ct. 1960 ). . Browning-Ferris Indus. v. Ter Maat, 195 F.3d 953 (7th Cir.1999) (Posner, J.). . Id. at 958 . . 42 U.S.C. § 9607 (a)(4)(B). . 40 C.F.R. § 300.160 (a)(1) (1998). . See Valley Eng'rs, Inc. v. Electric Eng’g Co., 158 F.3d 1051 , 1052 (9th Cir.1998) (stating that the district court’s findings of fact are reviewed for clear error); In re Bell Petroleum Servs., Inc., 3 F.3d 889, 896 (5th Cir.1993) (stating that actual apportionment of damages is a question of fact). . 42U.S.C. § 9613(f)(1). . See Bedford Affiliates v. Sills, 156 F.3d 416 , 423-24 (2d Cir.1998); Bancamerica Commercial Corp. v. Mosher Steel, 100 F.3d 792, 802 (10th Cir.1996); Environmental Transp. Sys. Inc., v. ENSCO, Inc., 969 F.2d 503 , 509 (7th Cir. 1992); United States v. R.W. Meyer, Inc., 932 F.2d 568, 571 (6th Cir.1991). . See In re Dant & Russell, Inc., 951 F.2d 246 , 249 (9th Cir.1991). . See Bell Petroleum, 3 F.3d at 896 (stating that actual apportionment of damages is a question of fact). . See, e.g., Acushnet Co. v. Mohasco Corp., 191 F.3d 69, 74 (1st Cir.1999); Centerior Serv. Co. v. Acme Scrap Iron & Metal Corp., 153 F.3d 344, 354 (6th Cir.1998); Control Data, 53 F.3d at 935 ; United States v. Colorado & Eastern R.R. Co.; 50 F.3d 1530 , 1536 n. 5 (10th Cir.1995); Kerr-McGee Chem. Corp. v. Lefton Iron & Metal Co., 14 F.3d 321 , 326 (7th Cir. 1994); Bell Petroleum, 3 F.3d at 899-900 . . 42 U.S.C. § 9613 (f)(1). . Bell Petroleum, 3 F.3d at 901 . . Id. at 903 . . Id. . See Bell Petroleum, 3 F.3d at 896 ; U.S. v. Alcan Aluminum, Corp., 990 F.2d 711 , 722 (2d Cir.1993). . Cree v. Flores, 157 F.3d 762, 768 (9th Cir. 1998) (quoting Anderson v. City of Bessemer City, 470 U.S. 564, 573 , 105 S.Ct. 1504 , 84 L.Ed.2d 518 (1985)). . Boeing Co. v. Cascade Corp., 920 F.Supp. 1121, 1140 (D.Or.1996). . Id. . 42 U.S.C. § 9613 (f)(2). . 42 U.S.C. § 9613 (f)(1). . 42 U.S.C. § 9614 (b). . See Fed.R.Civ.P. Rule 46; see also Jovanovich v. United States, 813 F.2d 1035, 1037 (9th Cir. 1987). . See Pinal Creek Group v. Newmont Mining Corp., 118 F.3d 1298, 1300 (9th Cir. 1997). . See 42 U.S.C. § 9613 (g). In relevant part, that provision states: In any such action described in this subsection [cost recovery actions], the court shall enter a declaratory judgment on liability for response costs or damages that will be binding on any subsequent action or actions to recover further response costs or damages. 42 U.S.C. § 9613 (g)(2). . See 42 U.S.C. § 9613 (g)(3). . See, e.g., Control Data, 53 F.3d at 936 . . See Wilton v. Seven Falls Co., 515 U.S. 277, 289 , 115 S.Ct. 2137 , 132 L.Ed.2d 214 (1995) ("We believe it more consistent with the statute to vest district courts with discretion in the first instance, because facts bearing on the usefulness of the declaratory judgment remedy, and the fitness of the case for resolution are peculiarly within their grasp.”); Government Employees Ins. Co. v. Dizol, 133 F.3d 1220, 1223 (9th Cir.1998) (stating that review of exercise of declaratory relief for abuse of discretion is appropriate where party made objection below and trial court issued written decision detailing its reasons). . Wickland Oil Terminals v. Asarco, Inc., 792 F.2d 887, 893 (9th Cir.1986) (quoting Maryland Cas. Co. v. Pacific Coal & Oil, Co., 312 U.S. 270, 273 , 61 S.Ct. 510 , 85 L.Ed. 826 (1941)) (citations omitted).