CLARK, Chief Judge:

The district court affirmed a magistrate's denial of motions to quash two administrative search warrants authorizing programmed safety inspections of steel plants by the Occupational Safety and Health Administration (OSHA). Finding that affidavits by local OSHA agents, OSHA programmed inspection instructions and statewide industry ranking reports provided probable cause to issue the warrants, we affirm.

I.

We are asked to decide whether the magistrates in these consolidated cases had probable cause to issue administrative search warrants under § 8(a) of the Occupational Safety and Health Act, 29 U.S.C. § 657(a) (1985) ("the Act"). The warrants, issued in ex parte proceedings as authorized by 29 C.F.R. § 1903.4(d) (1987), permitted programmed safety inspections of Industrial Steel Products Co.'s Shreveport steel fabrication plant and Mosher Steel Products Co.'s Houston steel plant to ensure compliance with the Occupational Safety and Health Act of 1970, 29 U.S.C. §§ 651-78 (1985).

Industrial and Mosher were selected for programmed safety inspections based on CPL 2.45A, OSHA's standard instructions for scheduling inspections of plants for which specific safety violations have not been reported. Local OSHA officials secured administrative search warrants after Industrial and Mosher refused to consent to the inspections. Having secured the warrants, OSHA officials again notified Industrial and Mosher that their plants would be inspected. Again they refused to consent and instead filed motions to quash the administrative inspection warrants. Separate hearings were held and federal magistrates denied both motions. Industrial and Mosher timely appealed the magistrates' findings of fact and recommendations. In orders dated June 8, 1987 and June 10, 1987, the district courts for the Southern District of Texas and the Western District of Louisiana adopted the magistrates' findings of fact and recommendations and denied the motions to quash. Industrial and Mosher appeal. We affirm.

II.

In reviewing a magistrate's determination of probable cause, this court may only consider material which the issuing magistrate had before him. Franks v. Delaware, 438 U.S. 154, 98 S.Ct. 2674, 2684, 57 L.Ed.2d 667 (1978); United States v. Melancon, 462 F.2d 82, 89-90 (5th Cir. 1972), cert. denied, 409 U.S. 1038, 93 S.Ct. 516, 34 L.Ed.2d 487 (1972). Evidence outside of the warrants may only be examined if the applications were tainted by fraud or misrepresentation. Franks, 98 S.Ct. at 2684. Because no allegation of fraud has been made, we will confine our review to the documents submitted with the warrant applications.

The Occupational Safety and Health Act authorizes the Secretary of Labor to enter and inspect work places at reasonable times and in a reasonable manner to ensure compliance with the provisions of 29 U.S.C. §§ 654(a)(1) and (2). In Marshall v. Barlow's, Inc., 436 U.S. 307, 98 S.Ct. 1816, 56 L.Ed.2d 305 (1978), the Supreme Court held § 657(a) unconstitutional insofar as it purported to authorize nonconsensual, warrantless inspections. Barlow's, 98 S.Ct. at 1827. However, the Court construed the Act to permit inspections pursuant to administrative search warrants which could be obtained upon a showing of: (1) "specific evidence of an existing violation," or (2) "`that reasonable legislative or administrative standards for conducting an ... inspection are satisfied with respect to a particular [establishment].'" Barlow's, 98 S.Ct. at 1824 ( quoting Camara v. Municipal Court, 387 U.S. 523, 87 S.Ct. 1727, 1736, 18 L.Ed.2d 930 (1967)).

In Brock v. Gretna Machine Ironworks, Inc., 769 F.2d 1110 (5th Cir. 1985), this Circuit considered the adequacy of OSHA's application for a warrant authorizing the entry of an establishment selected under OSHA's programmed health inspection plan. The panel in Gretna made its analysis in two parts. The first considered whether OSHA's health inspection plan met the specific neutral criteria mandated by the Supreme Court in Barlow's. The Gretna panel concluded it did.

In analyzing an inspection plan, the magistrate considering the warrant application must "review the plan itself to ensure that it contains the specific neutral criteria mandated by Barlow's." Gretna, 769 F.2d at 1112. Essentially, this part of the test requires the magistrate to determine that the plan as a whole is susceptible of neutral, nonarbitrary application.

We are satisfied that this prong was met in Industrial's and Mosher's case. The magistrates reviewed copies of CPL 2.45A and descriptions of the programmed inspection plans submitted with the warrant applications. Our examination of this same material discloses that CPL 2.45A and the inspection plans are based on specific, neutral criteria.

The plan states that the national OSHA office will provide each regional office with a statewide industry ranking report classifying all industries within the state under a Standard Industrial Classification (SIC) code. The SIC code is a four-digit number which classifies industries by what they manufacture (for example, wood partitions and fixtures or canned and cured seafoods). All industries within a given SIC code are then assigned a Lost Work Day Injury (LWDI) rate calculated by Bureau of Labor Statistics. The LWDI rate reflects the average number of work days per 100 full-time employees lost by employees in that industry as a result of occupational injury or sickness.

To target specific businesses within a state for inspection, OSHA's national office provides each regional office with establishment lists stating the names and addresses of all businesses in the state in each SIC code on the statewide industry ranking report. There is one establishment list for high-hazard industries (defined as industries with an LWDI rate above the national average of 3.4), one for low-hazard industries, and one for nonmanufacturing industries. The national office obtains these names and addresses from Dun's Marketing Service. The establishment lists rank industries in descending order starting with the industry in the SIC codes having the highest LWDI rate. Within each SIC code, businesses are separated by county and then placed in alphabetical order.

Using specific criteria supplied by OSHA's area director, the regional office then adds or deletes establishments from the high-hazard establishment list. An establishment may only be added if it has more than ten employees but the national office thought it had fewer than ten, or if it was unlisted but is believed to be within an SIC code on the statewide industry ranking report. An establishment may only be deleted if: it cannot be located, is no longer in business, is out of area office jurisdiction, is a nonplant corporate office, has been listed under an incorrect SIC code and the correct SIC code is not on the statewide industry ranking report, a safety inspection has been conducted within the current or previous two fiscal years, it has ten or fewer employees, or its actual LWDI rate is found to be lower than the national average during an OSHA health inspection. Deletions may be made for other reasons only upon the approval of OSHA's regional administrator and director of field operations. After additions and deletions have been made, establishments on the list are numbered consecutively.

Next, each regional office compiles an inspection register naming establishments within its jurisdiction which will be inspected during the current fiscal year. The number of establishments on the register is determined by doubling the number of programmed inspections projected for the fiscal year (to account for establishments later found to be exempt) and subtracting any uncompleted inspections carried over from the previous fiscal year. Ninety percent of that number is taken in rank order from the top of the high-hazard establishment list. Five percent is selected randomly from the low-hazard establishment list and five percent from the nonmanufacturing list. The random selections give low-hazard and nonmanufacturing businesses incentive to comply with OSHA guidelines. The register is then divided into two equal cycles. With certain limited exceptions, each establishment in the first cycle (the "worst" establishments) must be inspected before the second cycle is begun. Within each cycle, establishments may be inspected in the order that makes the most efficient use of OSHA's resources. Establishments on the register will be exempt from a comprehensive inspection if a preliminary review of their OSHA logs and injury records reveals a lower than average LWDI rate. However, every "tenth establishment" will receive a full-scale inspection regardless of its actual LWDI rate. This assures that even if businesses falsify their injury records, they still may be subject to inspection.

This plan is based on specific, neutral criteria. Establishments employing the most workers and having the highest LWDI rates are subject to the most frequent inspections. OSHA's assertion that it has insufficient resources to inspect every establishment is undisputed. Thus, it is reasonable for OSHA to concentrate on the largest, most dangerous businesses. The addition of some low-hazard and nonmanufacturing firms to the register and the comprehensive inspection of some firms with low actual LWDI rates enhances the plan's neutrality. These devices further OSHA's legitimate goals of encouraging all firms to comply with its regulations and discouraging firms with high LWDI rates from manipulating accident statistics to avoid comprehensive inspections.

Inspecting firms within a cycle in the order that makes the most efficient use of OSHA's resources casts no shadow on the plan's neutrality. Establishments from the top of the high-hazard list will be in the first cycle and will all be inspected before inspection of the lower-hazard firms in the second cycle begins. In an inspection scheme as pervasive as this one, maintaining precise order is neither always possible nor essential to neutrality. Subjecting a high-hazard firm with a slightly lower LWDI rate to inspection before a higher-hazard firm is not arbitrary. See, e.g., Donovan v. Trinity Industries, Inc., 824 F.2d 634, 636 (8th Cir. 1987). All firms within a cycle will be inspected in the space of several months in any case. Rearranging their order within the cycle is not discriminatory as Industrial and Mosher contend. Because it furthers OSHA's legitimate goal of efficient resource allocation, it is reasonable.

Industrial argues that the plan is not neutral because firms on the top of the high-hazard establishment list are repeatedly inspected while firms on the bottom of the list may never be inspected. By way of illustration, Industrial states that there are approximately 1400 firms on the Louisiana high-hazard establishment list. If the 300 firms from the top of the list are placed on the register each year, and are only exempt if they've been inspected in the past two years, then the 500 firms on the bottom of the list will never be subject to inspection.

Industrial's argument assumes that the establishment lists are unvarying from year to year. This is not the case. Indeed, the whole purpose of OSHA inspections is to cause LWDI rates to change for the better.

Businesses move up and down the establishment list each year as the LWDI rate for their SIC code changes. Mosher Steel is a prime example. The LWDI rate for SIC Code 3441 establishments in Texas was 7.0 in 1985 and had changed to 8.9 by 1986. Hence, it is not true that firms on the top of the establishment list will remain at the top and be subject to repeated inspections. Most high-hazard businesses will eventually be fair game for inspection as they move up or down the list. With this in mind, it is legitimate for OSHA to single out for inspection firms appearing at the top of the high-hazard list in a given fiscal year.

III.

Gretna also required the magistrate be able to determine that the business opposing the warrant was appropriately selected for inspection under the plan's neutral criteria. This determination requires the magistrate's "consideration of the plan's industry rank list for the state in question and the methodology used in the selection of the particular establishment." Gretna, 769 F.2d at 1112. Gretna goes on to state that the magistrate "should also be informed of the Secretary's perception of the desired frequency of inspections of companies on the establishment list and the actual inspection history of the business in question." Gretna, 769 F.2d at 1112. Because the warrant applications in today's cases supplied this information, it is unnecessary for us to decide whether this precatory language should be read as essential to any finding of probable cause.

The warrant applications before us contain more than enough information to support the magistrate's findings that OSHA had satisfied the second part of Gretna. Each application appended a copy of the applicable statewide industry ranking reports. Affidavits submitted with the warrants reviewed the methodology used to select Industrial and Mosher and appended copies of CPL 2.45A. One affidavit was sworn to by a local OSHA compliance officer and the other by an OSHA safety supervisor.

Industrial and Mosher argue that the affidavits failed to establish probable cause because they were not based on personal knowledge. Giordenello v. United States, 357 U.S. 480, 78 S.Ct. 1245, 1250, 2 L.Ed.2d 1503 (1958). They contend that not only did the affiants fail to allege personal knowledge, but, in fact, they did not have personal knowledge because they did not know how the establishment lists were compiled or what local additions or deletions had been made to the lists. Additionally, one affiant was not involved in scheduling inspections and did not personally review the statewide industry ranking report, the inspection register or the establishment list.

Industrial and Mosher are giving an unduly technical and restrictive reading to the affidavits. United States v. Ventresca, 380 U.S. 102, 85 S.Ct. 741, 747, 13 L.Ed.2d 684 (1965). One affiant, a compliance officer, regularly conducted OSHA inspections. The other, a safety supervisor, was responsible for enforcing the Act in 26 Texas counties and for scheduling inspections to be conducted by compliance officers. Both have gained sufficient personal knowledge in the course of these duties to testify regarding the procedures used to select establishments for inspection. Additionally, the affiants are entitled to rely on the actions and observations of their fellow OSHA personnel to supplement their own personal knowledge. Ventresca, 85 S.Ct. at 747; United States v. Flynn, 664 F.2d 1296, 1303 (5th Cir. 1982), cert. denied, 456 U.S. 930, 102 S.Ct. 1979, 72 L.Ed.2d 446 (1982).

The Secretary argues that Mosher is collaterally estopped from arguing lack of personal knowledge because the issue was previously litigated and decided adversely to its parent corporation, Trinity Industries. See, Donovan v. Trinity Industries, Inc., 824 F.2d 634, 636 (8th Cir. 1982). However, the Secretary waived the affirmative defense of collateral estoppel by failing to raise it before the district court. Mozingo v. Correct Manufacturing Corp., 752 F.2d 168, 172 (5th Cir. 1985).

Each affidavit contains a detailed description of how the plan was executed. These affidavits contain ample information to enable the magistrates to decide that Industrial and Mosher were appropriately chosen by an application of the plan's neutral criteria.

Information regarding the Secretary's perception of the desired frequency of inspections also was available for the magistrates' consideration. The affidavits state that, pursuant to the plan, an establishment will be exempt if it has undergone a programmed or unprogrammed safety inspection within the previous two fiscal years or the current fiscal year. The plan indicates that high-hazard establishments should be inspected for safety no more often than once every two to three years. Both Industrial and Mosher have been inspected somewhat more frequently. However, neither has been inspected within the previous two years.

Mosher contends that it has not been inspected in accordance with the Secretary's desired frequency because the affidavit misrepresented its actual inspection history. However, even under Mosher's version of its history, it would not qualify for an exemption by reason of inspection in the previous two fiscal years or the current fiscal year.

The affidavits and plan enabled the magistrate to determine that Industrial and Mosher were selected for inspection through an application of the plan's specific, neutral criteria. However, Industrial and Mosher argue that Gretna imposes an additional requirement — that the relevant establishment lists and/or the inspection register be appended to the warrant applications. Industrial and Mosher find support for this argument in a portion of Gretna which reads:

"But the application contained neither the establishment list nor a description of the procedure followed in the selection of Gretna.

....

... Indeed, OSHA provided the court in this case with an industry rank list from Connecticut and an establishment list from a county in Massachusetts. We see no reason why OSHA could not have provided the court with the relevant lists from Louisiana, encoded to protect the identity of other inspection targets."

769 F.2d at 1113 and n. 6.

Industrial and Mosher interpret this language to require that encoded establishment lists be appended to every warrant application. They argue that without encoded establishment lists or inspection registers, the magistrate cannot determine if local additions and deletions have been properly made to the list, if the list is in worst-first order, if it is complete and if it was compiled by a correct application of the neutral criteria. Industrial and Mosher state that a simple description of how the establishment lists were compiled is not adequate because without the lists themselves, the magistrate can't ascertain that the neutral procedures were actually used.

A footnote in Gretna observed that local additions and deletions are a matter for concern because "[o]therwise, OSHA officers could harass businesses or effectively exercise the kind of unbridled discretion which concerned the Supreme Court in Barlow's." Gretna, 769 F.2d at 1113, n. 7. However, they are not a problem in this case because the affidavits alleged that only local additions and deletions "appropriate" under CPL 2.45A were made. In the absence of evidence to the contrary, the magistrates are entitled to rely on these representations. Neither Industrial nor Mosher attempted to show that they were inappropriately added at the local level because they were both on the lists supplied by the national office.

Industrial contends that many of its competitors were inappropriately deleted from the establishment list. It cites statistics from the Louisiana Department of Commerce placing 93 businesses in Louisiana in SIC Code 3441. OSHA's Statewide Industry Rank Report for Louisiana only listed 42 businesses in Louisiana under SIC Code 3441. However, because these `deletions' did not occur at the local level, they do not raise the concerns expressed in Gretna. Nor can we conclude from anything shown by Industrial that the national office arbitrarily deleted Industrial's competitors from the list. OSHA applies its programmed inspection plan nationwide and therefore uses national directories classifying establishments by SIC code. OSHA couldn't feasibly rely on state directories because not all states publish them, some classify industries using different criteria than the national directory, some states use their directories for advertising purposes so businesses represent that they deal in more than one industry and some states publishing directories don't require that businesses include information in them.

Gretna does not mandate that an encoded establishment list or inspection register be adduced in the application for an administrative warrant in order to show probable cause. Local officials have the option of supplying either an encoded establishment list or the register; or, a description of how the list and register were compiled. The test is whether the magistrate can be satisfied that the target establishment was placed on the register by an application of the plan's specific, neutral criteria. The absence of an encoded establishment list will only be fatal to a warrant application where the target company can make a factual showing that it was placed on the list for reasons other than the application of specific, neutral criteria.

Industrial's and Mosher's suggestion that the magistrate should review the establishment list for completeness and neutrality would, in effect, require the magistrate to determine that every firm on the list was fairly placed there and that no firm was unfairly deleted. The Fourth Amendment does not impose such stringent requirements. To meet the standards set by Barlow's, the magistrate need only conclude that an application of the neutral plan resulted in the selection of the individual establishment contesting the warrant. Absent a specific factual showing by the firm contesting enforcement of the warrant, OSHA is not required to demonstrate that the plan was applied neutrally to every establishment on the list.

If a contestant puts forward proof that businesses were arbitrarily added to or deleted from the list, the court can no longer rely on the affiant's representations that a specific neutral plan had been properly applied. Such a showing would require the court to go behind the affidavits and examine the encoded establishment lists to verify the evidence that arbitrary additions or deletions had been made. Industrial and Mosher have made no such showing of arbitrariness in this case. They have merely suggested that capricious additions and deletions could have been made at the local level. That is not enough.

IV.

The district courts' orders denying Industrial's and Mosher's motions to quash are

AFFIRMED.