OPINION OF THE COURT
SLOVITER, Circuit Judge.
Before us is a petition for review by Hunter Douglas, Inc. from a decision and order of the National Labor Relations Board and a cross-application by the Board for enforcement of its order. The portion of the order contested by Hunter Douglas involves the Board’s determination that Hunter Douglas’ discharge of most of the second-shift employees at its Maywood, New Jersey plant violated sections 8(a)(1) and (3) of the Labor Management Relations Act (the “Act"), 29 U.S.C. §§ 158(a)(1), (3), and that Hunter Douglas’ questioning of employee Victor Nunez violated section 8(a)(1) of the Act, 29 U.S.C. § 158(a)(1).
I.
Facts
Hunter Douglas manufactures window coverings and distributes both components for window coverings and window covering manufacturing equipment to fabricators and assembles the coverings itself. In the [810]*810United States, it operates plants in May-wood, New Jersey; Dallas, Texas; Ontario, California; and Kent, Washington. The present case involves its actions at the Maywood plant.
During the summer of 1984 that plant operated on two shifts, with 63 employees on the first shift and 42 employees on the second. In August 1984, an employee informed Jose Lugo, the business agent of Local 404 of the United Electrical, Radio & Machine Workers of America, that May-wood employees were seeking to form a union. Subsequently, there were four meetings of plant employees at the union’s offices. The first, on August 29, 1984, was attended by six employees. Later meetings on September 15 and 29 and October 20 were attended by small but increasing numbers of employees. During this period, employees were also distributing union leaflets at the plant. At the fourth meeting, union authorization cards were distributed, and, by October 24, 1984, 73 signed authorization cards had been obtained, 36 from first-shift employees and 37 from second-shift employees.
In late August or early September 1984, John Santalla, the Maywood plant manager, received an anonymous letter mentioning the names of several employees who were trying to organize a union. Santalla then created employee committees, which he met with on September 18, 20 and 27 and October 5 and 17 in the plant cafeteria on working time. At the meetings, employees complained of working conditions, such as problems with the heating system, air conditioning, bathrooms, and cafeteria. Santalla said he would try to resolve these problems. The employees also mentioned the possibility of receiving more pay. Santalla replied that the company was trying to initiate an incentive pay plan. Medical insurance was also discussed. At the third meeting, Santalla asked each of the employees whether their previous employer had had a union, to which some responded that it had and others that it had not. At the fourth meeting, Santalla distributed a list of proposed shop rules for the committee’s consideration. One of the rules prohibited solicitation of employees if either the soliciting or the solicited employee was on working time. Another of the rules prohibited distribution of handbills and literature in plant working areas.
In October 1984, Santalla asked a machine operator, Victor Nunez, who went to Santalla’s office to get a band-aid, whether it was true that Nunez wanted to join the union. Nunez replied “John, you know that I promised you that I do not want to know anything about unions, and I am not going to join any unions.” App. at 683. Santalla then told Nunez no one else was to know about the conversation. A second conversation occurred in the area of the restrooms. Santalla said that he wanted to ask Nunez a confidential question and inquired whether it was true that the employees “were trying to get the Union back in place.” App. at 684. Nunez replied, “I promised you that I wasn’t going to get into a union, and I don’t want to talk about it anymore.” App. at 684.
On October 24, 1984, Hunter Douglas discharged 35 of the 42 second-shift employees and the second-shift supervisor. It is the General Counsel’s position that the company’s motive in discharging the second-shift employees was to inhibit union activity.
Hunter Douglas, on the other hand, argues that these firings resulted from legitimate business reasons stemming from its strategy to cease all work for retail customers. Even before the events at issue, Hunter Douglas had sold its products primarily through independent fabricators. For approximately 90 percent of its window coverings, Hunter Douglas would produce the components but then distribute them to its network of fifty to sixty fabricators who were responsible for assembly and retail sales. In the spring and summer of 1984, Hunter Douglas decided to direct all orders through its fabricator network. In August 1984, it advised retail customers that after August 31, 1984 all orders should be directed to its fabricators.
[811]*811Hunter Douglas claims that it realized this strategy would give it excess production capacity, and that in spring 1984 it began to plan ways to reduce that capacity. Between March and July 1984, it reduced management personnel by 50%. It made similar cuts in space requirements and inventory during summer and fall 1984. Hunter Douglas claims that it would have had reduced manufacturing manpower needs at the Maywood plant in early September 1984, but that it was able to stave off a long-planned reduction in manufacturing workers at the Maywood plant because of orders from J.C. Penney, one of its largest fabricators. Penney had oversold its capacity and sought production assistance from Hunter Douglas the first week in September.
On October 18,1984, O.B. Kelley, Hunter Douglas’ Vice President of Sales and Marketing, learned that Penney would place no more orders with Hunter Douglas. Kelley informed George Shouldis, Vice President and General Manager of Hunter Douglas. On October 23,1984, the loss of the Penney orders was confirmed. That same day Shouldis instructed John Brown, Director of Manufacturing, to develop a plan to reduce manpower. That same day, Brown, after consultation with Santalla, recommended to Shouldis that the second shift be eliminated. On the next day, October 24, 1984, with no notice, 35 of the Maywood plant’s 42 second-shift employees and the second-shift supervisor, Jose Algarin, were informed of their discharge. Hunter Douglas explains that the dismissal without notice was effected to spare its employees from feelings of “unrest” and “discomfort”. App. at 795. Local 404 formally demanded recognition from Hunter Douglas on October 29, 1984.
II.
Proceedings
Local 404 filed three unfair labor practice charges against Hunter Douglas, which were consolidated in a complaint issued by the Regional Director. Following a hearing, the Administrative Law Judge found, inter alia: that Hunter Douglas had violated sections 8(a)(1) and (2) of the Act by establishing committees, soliciting grievances, granting benefits and promulgating prohibitions on solicitation and distribution, all as a means of discouraging union organization; and that there was not sufficient evidence to show that Algarin, the second-shift supervisor, had been discharged for his refusal to commit unfair labor practices.
With respect to the activities at issue here, the AU dismissed the allegation that elimination of the second shift was an unfair labor practice. The AU found: “Respondent [Hunter Douglas] had knowledge of the Union’s activities, Santalla displayed animus towards the Union and the timing of the discharges was such that they occurred during the Union campaign.” App. at 40. He stated that although 90% of the second-shift employees had signed authorization cards, the discharges occurred before Hunter Douglas was aware of the signing of the cards and before Local 404 demanded recognition. The AU stated it was “questionable”, but did not decide whether the General Counsel made a prima facie showing that protected conduct was a motivating factor in the decision to eliminate the second shift. App. at 40. He found no violation had occurred because he concluded that Hunter Douglas would have eliminated the second shift even in the absence of the protected conduct.
The AU also concluded that the questioning of Victor Nunez about union activities was not coercive under the totality of the circumstances and therefore was not an unfair labor practice. Noting that Nunez was not summoned to Santalla’s office on the first occasion and that the two met casually near the restrooms on the second occasion, the AU found that Santalla’s inquiries were the kind of “casual questioning concerning union sympathies” permitted by the Act. App. at 39 (quoting Graham Architectural Products Corp. v. NLRB, 697 F.2d 534, 541 (3d Cir.1983)).
Only the General Counsel filed exceptions to the AU’s decision. The Board [812]*812affirmed the AU’s rulings, findings, and conclusions except with respect to the discharge of the second shift and the interrogation of Nunez. Hunter Douglas, 277 NLRB No. 123, slip op. at 7 (Dec. 23, 1985). The Board held that the ALJ’s analysis of the discharge of the second shift was faulty because it failed “to consider record evidence that demonstrates that the second shift was selected for layoff in order to defeat the union campaign.” Id. at 3. Drawing different inferences from the facts on record, the Board concluded that the General Counsel made a prima facie case that union animus was a motivating factor in the layoff of second-shift employees, and that Hunter Douglas failed to show by a preponderance of the evidence that those employees would have been laid off in the absence of union activity and animus. Id. at 7. It thus held that the discharge of its second-shift employees violated sections 8(a)(1) and (3) of the Act.
The Board also rejected the AU’s conclusion that Santalla’s questioning of Nunez was not coercive. The Board stressed that Santalla told Nunez the inquiries were confidential, that the ■ questions had no legitimate purpose, and that Santalla was a plant manager who had conducted the interrogation in a calculated fashion. Id. at 6-7. The Board, therefore, held that Santalla’s questioning of Victor Nunez on two occasions violated Section 8(a)(1) of the Act because it tended to interfere with the exercise of employees’ protected rights. Id. at 7.
III.
Standard of Review
We must sustain the Board’s findings of unfair labor practices if they are supported by substantial evidence on the record considered as a whole. Universal Camera Corp. v. NLRB, 340 U.S. 474, 487-91, 71 S.Ct. 456, 463-66, 95 L.Ed. 456 (1951); Graham Architectural Products Corp. v. NLRB, 697 F.2d 534, 537 (3d Cir.1983). Similar deference is granted the Board’s inferences from facts which, if supported by substantial evidence, may not be displaced even though this court acting de novo might have reached a different conclusion. Universal Camera, 340 U.S. at 488, S.Ct. at 464; Herman Brothers, Inc. v. NLRB, 658 F.2d 201, 209 (3d Cir.1981); NLRB v. Garry Manufacturing Co., 630 F.2d 934, 937 (3d Cir.1980).
Here, the Board disagreed with the AU. Hunter Douglas contends that this should subject the Board’s findings to stricter scrutiny by this court. See Eastern Engineering & Elevator Co. v. NLRB, 637 F.2d 191, 197 (3d Cir.1980). See also Albritton Communications Co. v. NLRB, 766 F.2d 812, 817 (3d Cir.1985), cert. denied, — U.S. —, 106 S.Ct. 850, 88 L.Ed.2d 891 (1986) (“When, as here, the Board rejects certain findings made by an AU, this ‘substantial evidence’ standard is cast in a special light”). The Board, on the other hand, argues that when the differences between the Board and the AU stem solely from different inferences from the overall record evidence, we are to attach no special weight to the AU’s contrary conclusions, since it is the Board’s decision that is under review. See Stein Seal Co. v. NLRB, 605 F.2d 703, 706 (3d Cir.1979); NLRB v. Duquesne Electric & Manufacturing Co., 518 F.2d 701, 704-05 (3d Cir.1975).
. This issue was discussed in our in banc decision in Hedstrom Co. v. NLRB, 629 F.2d 305 (3d Cir.1980), cert. denied, 450 U.S. 996, 101 S.Ct. 1699, 68 L.Ed.2d 196 (1981), where we stated:
[T]he Board is not in any way inhibited by inferences drawn by the AU. To the contrary, the authority to draw legitimate inferences from proven facts is exclusively the Board’s, not the AU’s or the court on review of the Board’s order. Indeed, we have consistently held that “the Board has the power to draw different conclusions from evidentiary facts” presented to the AU, including conclusions that directly contradict those reached by the AU. (citations omitted). If more than one inference may be drawn from a given set of facts, therefore, the [813]*813conclusion of the Board will control unless it is unreasonable.
Id. at 316.
In this case, the Board did not make credibility findings that differed from the ALJ. Instead, the Board exercised its authority to draw “legitimate inferences from proven facts.” We must therefore review its findings under the generally applicable substantial evidence standard. See Garrett Railroad Car & Equipment, Inc. v. NLRB, 683 F.2d 731, 739-40 (3d Cir.1982).
IV.
Discharge of the Second-Shift Employees
A. Prima Facie Case
In determining whether a discharge of employees constitutes an unfair labor practice under sections 8(a)(1) and (3) of the Act, the General Counsel bears the initial burden of making a prima facie showing by a preponderance of the evidence that protected conduct was a motivating factor in the employer’s conduct. Once the General Counsel demonstrates a prima fade case, the burden shifts to the employer to demonstrate that it would have taken the same action even in the absence of protected conduct by the employees. See NLRB v. Transportation Management Corp., 462 U.S. 393,103 S.Ct. 2469, 76 L.Ed.2d 667 (1983); Hanlon & Wilson Co. v. NLRB, 738 F.2d 606, 610 (3d Cir.1984); Wright Line, a Division of Wright Line, Inc., 251 N.L.R.B. 1083 (1980), enforced, 662 F.2d 899 (1st Cir.1981), cert. denied, 455 U.S. 989, 102 S.Ct. 1612, 71 L.Ed.2d 848 (1982).
Although the AU made no specific finding whether a prima facie case had been made out, the Board explicitly stated as a conclusion of law that the General Counsel had established a prima facie case. Hunter Douglas argues that there is no substantial evidence to support this conclusion. In essence, its argument is that a prima facie case depends on a showing that the discharge of the second-shift employees was motivated by antiunion animus. It argues that such a showing was not made because there was no evidence that Shouldis or Brown, the two company officers responsible for the elimination decision, had any knowledge of union activity or harbored antiunion animus towards that activity. Hunter Douglas points to the fact that the first demand for recognition by Local 404 did not occur until October 29, 1984, five days after the second shift’s elimination, and it claims that until that demand, union activity was so minimal as to be unnoticeable.
Concededly, attendance at union meetings held prior to the second shift’s elimination was not large. The testimony was that six workers attended the first meeting, somewhat more attended the next two meetings, and 11 attended the fourth meeting. Meetings were not, however, the sole extent of union activity. Union literature in both English and Spanish, including 200 leaflets, was distributed in and around the plant in September and October. On October 20, the distribution of union authorization cards began. By October 24, approximately 50% of the workers on the first and second shifts had signed and returned authorization cards.
Hunter Douglas does not contend that plant manager Santalla was unaware of the union activity. His keen interest in that activity, as well as his antiunion animus, is fully substantiated on the record. In September, Santalla questioned second-shift supervisor Algarin about possible union activity by an employee. In September and October, Santalla organized and participated in the employee committee meetings at which the no-solicitation and no-distribution rules which would have precluded union literature were promulgated. After one meeting, Santalla questioned Algarin about an employee who had complained about low pay and Santalla told Algarin, “I don’t care what you do, get rid of her. If anybody has anything to do with the Union, she’s one of them,” App. at 521; see also App. at 957-58. Santalla questioned employee Nunez about his union activity twice in October. Most significantly, in October, Santalla instructed Algarin to watch for [814]*814union activists in the second shift as Santalla wanted to “get the teeth out of the union.” App. at 528.
Hunter Douglas points to direct testimony by Brown and Shouldis that they had no knowledge of union activity. However, Brown also testified that he was always “aware there was a potential for union activity. And from time to time I heard comments, but nothing that I could recall specifically.” App. at 860. The ALJ did not find that Brown and Shouldis had no knowledge or union activity. He did find, without further elaboration, that “Respondent [Hunter Douglas] had knowledge of the Union’s activities.” App. at 40.
It is significant that when Shouldis asked Brown on October 23 to formulate a personnel reduction plan, Brown then consulted with Santalla, who was well aware of the increasing union activity. Santalla admitted in his testimony that he recommended to Brown that the second shift be eliminated. App. at 976. Brown in turn relied on Santalla’s recommendation in suggesting the second shift’s elimination to Shouldis. App. at 878. Where, as here, a supervisory employee has knowledge of union activity, an inference that he has communicated that knowledge to his superiors is permissible. See Texas Aluminum Co. v. NLRB, 435 F.2d 917, 919 (5th Cir.1970).
In some cases, the courts have upheld the Board’s inferences of knowledge of union activity solely on the basis of the timing and nature of the discharge despite company claims that no management officials were aware of union activity. See NLRB v. Wal-Mart Stores, Inc., 488 F.2d 114, 117 (8th Cir.1973); NLRB v. Long Island Airport Limousine Service Corp., 468 F.2d 292, 295-96 (2d Cir.1972); NLRB v. Sequoyah Mills, Inc., 409 F.2d 606, 609 (10th Cir.1969). In this case as well, a similar inference is permissible, not only from the plant manager’s knowledge but from the timing and nature of the discharge.
The company’s abruptness in implementing Brown’s recommendation is significant. Employees were given no notice of their discharge and were simply informed when they arrived at work on October 24 that they no longer had jobs. This abrupt discharge coincided with the culmination of Local 404’s drive to obtain union authorization cards. The only explanation offered for the lack of any notice prior to discharge, Shouldis’ testimony that, “I think decisions as dramatic as this require immediate implementation. I think to drag along only causes unrest and discomfort for all concerned,” App. at 795, is implausible.
Moreover, the manner of implementing the abrupt layoff departed from the company’s past practice of reassigning employees to other shifts. When Hunter Douglas eliminated its third shift in 1983, it used the opportunity to “weed out some of the deficien[t]” employees, by reassigning others to the first two shifts. App. at 797. In 1984, a layoff was spread through the first and second shifts, using the same weeding out procedure. App. at 796-97. Although Hunter Douglas retained some second-shift employees with special skills, it has never suggested that it employed the same type of weeding out process that it had previously employed, and instead discharged almost all of the 42 second-shift employees in one fell move. The company cannot rationalize its action at the Maywood plant by its subsequent decisions to eliminate the second shift at the Kent, Washington plant in October 1984 and at the Ontario, California plant in January 1985.
Hunter Douglas contends that other evidence dispels any inference of discriminatory motive in the second-shift’s elimination. It argues that because almost equal numbers of union cards were obtained from each shift, 36 from the first shift and 37 from the second shift, both shifts had relatively equal amounts of union activity. By looking only at the numbers, Hunter Douglas ignores the relevant percentages. Because the first shift had 63 employees while the second shift had only 42, only 56% of the first shift but over 90% of the second shift signed union cards. Moreover, the second-shift employees were more [815]*815activé in engaging in union activity. All six employees who attended the initial union meeting were from the second shift as were a majority of attendees at each of the subsequent meetings. Second-shift employees were primarily involved in the distribution of literature and union cards. Most important, Santalla’s activities were concentrated on the second shift. Second-shift supervisor Algarin testified that Santalla said “that he was aware that some of the employees were trying to organize a union and that he believed that it was coming from [the second] shift.” App. at 527.
Hunter Douglas also emphasizes that it retained ten union supporters from the second shift. Even if true, this does not change the fact that the majority of the second shift, which was the focus of union activity, was laid off. Moreover, courts have held that it is “irrelevant that some union sympathizers were not laid off, when the layoffs were meant to chill union activity,” NLRB v. Rain-Ware, Inc., 732 F.2d 1349, 1355 (7th Cir.1984). See also Birch Run Welding & Fabricating, Inc. v. NLRB, 761 F.2d 1175 (6th Cir.1985).
We have previously stated that “because there often is no direct evidence of antiunion motivation, the Board may rely on circumstantial evidence to prove such intent.” NLRB v. Scott Printing Corp., 612 F.2d 783, 787 (3d Cir.1979). See also NLRB v. Treasure Lake, Inc., 453 F.2d 202, 205 (3d Cir.1971). In this case, there was sufficient circumstantial evidence to support the reasonable inferences made by the Board. Even if Brown and Shouldis were not personally aware of the concentrated union activity that had taken place in the plant in the two months before elimination of the shift, since Santalla had both knowledge of that activity and antiunion animus and his recommendation played a part in the ultimate decision, Hunter Douglas may be held to have acted out of anti-union animus. Otherwise, companies will be able to accomplish impermissibly motivated discharges without penalty through the simple expediency of dividing their personnel functions and insulating top management from common knowledge. See Allegheny Pepsi-Cola Bottling Co. v. NLRB, 312 F.2d 529, 531 (3d Cir.1962). Accordingly, we hold that there was substantial evidence to support the Board’s conclusion that the General Counsel had made out a prima facie case.1
B. The Company’s Articulated Business Reason
Hunter Douglas also argues that even if the General Counsel established a prima facie case, the ALT correctly found that Hunter Douglas proved that the discharge of the second shift would have occurred even in the absence of any union organizing conduct. The company bases this claim on evidence of its overall policy of work reduction and reorganization, the unexpected termination of orders by J.C. Penney, and its “sharply” reduced production needs after the layoff.
The Board did not reject Hunter Douglas’ position that some form of work reduction would have occurred because of the change in its marketing strategy. It concluded, however, that Hunter Douglas failed to sustain its burden of proving that the timing and nature of the reduction that did occur resulted from business reasons, as it claimed, rather than from an attempt [816]*816to discourage unionization. See Transportation Management, 462 U.S. at 400-03, 103 S.Ct. at 2473-75.
In reaching this conclusion, the Board relied on the “abruptness of the layoff, coming as it did at the time cf. the union drive,” and the fact that “this layoff deviated from previous layoffs, in that it involved elimination of an entire shift rather than a reduction in the number of overall employees based on individual considerations.” Hunter Douglas, 277 NLRB No. 123, slip op. at 5-6.
The mass of economic data that the company lays before us does not convince us that there is not substantial evidence to support the Board’s decision. Hunter Douglas claims it had sharply reduced production needs, but in fact it took some time for production to decrease. The company’s average incoming orders for the first three weeks of November were substantially equivalent to the company’s average in October. Although the number of orders then declined for the two weeks near the Thanksgiving holiday, in December they returned to their approximate October level. App. at 290-91. The plant continued to be busy after the layoff, and the remaining first-shift workers were given the opportunity of working overtime “[j]ust about every day” until two to three weeks before Christmas. App. at 636-37. This tends to refute Hunter Douglas’ argument that the loss of J.C. Penney orders required immediate reduction of the workforce. In fact, in its opinion the Board noted that Shouldis testified that Penney’s termination of orders “was not a loss” because Hunter Douglas did not rely on Penney’s unsolicited business, and that Brown testified that the layoff was not “precipitated by any one event.” Hunter Douglas, 27 NLRB No. 123, slip op. at 4; see also App. at 821, 847-48.
The Board was not obliged to credit Hunter Douglas’ contention that lower productivity on the second shift was the cause of its elimination, particularly in light of the layoff of workers who had been commended by Santalla for their productivity and one worker who had participated in training of third-shift employees. App. at 454, 501-02. Based on the record that had been adduced before the AU, and without overturning any of his credibility determinations, the Board concluded that “the layoff of second-shift employees would not have happened as it did were it not for the desire of the Respondent to impede union activity.” Hunter Douglas, 27 NLRB No. 123, slip op. at 6. There was substantial evidence to support that determination.
V.
Interrogation of Nunez
Hunter Douglas also challenges the Board’s determination that Santalla’s interrogation of employee Victor Nunez tended to interfere with the exercise of employees’ protected rights and therefore was in violation of Section 8(a)(1) of the Act. In Graham Architectural Products, Corp. v. NLRB, 697 F.2d 534, this court stated:
An employer’s questioning becomes coercive and runs afoul of section 8(a)(1) when it “suggests to the employees that the employer may take action against them because of their pro-Union sympathies ____” Although the Board need not show that the employer’s interrogation actually had any coercive effect, the questioning must reasonably have tended to coerce under the circumstances.
Id. at 537 (quoting Frito-Lay, Inc. v. NLRB, 585 F.2d 62, 65 (3d Cir.1978)). See also NLRB v. Keystone Pretzel Bakery, Inc., 696 F.2d 257, 259-60 (3d Cir.1982) (in banc); Hedstrom Co. v. NLRB, 629 F.2d 305, 314.
Here, Santalla questioned Nunez during the first week of October when Nunez had gone to Santalla’s office to get a band-aid. Nunez testified that Santalla
wanted to know if it was true that I wanted to get into the Union again. I said, “John, you know that I promised you that I do not want to know anything about unions, and I am not going to join any unions.” He said ... that matter was [817]*817something between he and I, and that no one else was to know about it.
App. at 683. Although Hunter Douglas argues that this was the kind of “casual questioning concerning union sympathies” which is permissible under the Act, see Graham Architectural Products, 697 F.2d at 541, the Board noted Santalla’s direction that the conversation be kept confidential, which suggests to us that the questioning was more than “casual”. Although Nunez’ answer implies a promise made and no need to discuss the matter again, Santalla did question Nunez again, this time near the restrooms. Nunez testified that Santalla:
told me that he wanted to [ask] a confidential question of me. That he wanted to know if it was true that we were trying to get the Union back in the place. I told him — I promised him — “I promised you that I wasn’t going to get into a union, and I don’t want to talk about it anymore.”
App. at 684. •
Such repeated confidential questioning was unlikely to be taken as “casual” by the employee. As the Board pointed out, Santalla had no legitimate purpose for the questions. We find substantial evidence to support the Board’s findings regarding the interrogation in the nature of the questions, their repetition, their lack of any legitimate purpose, Santalla’s high management position, and his emphasis on their confidentiality.
VI.
Conclusion
Hunter Douglas does not oppose the Board’s request for enforcement of the portion of its order based on its findings that the company violated Sections 8(a)(1) and (2) of the Act by forming employee committees and convening meetings of the committees to solicit and resolve grievances, and that the company violated Section 8(a)(1) by granting benefits, remedying grievances, and promulgating a no-solicitation and no-distribution rule. Because we have found that substantial evidence supports the Board’s other findings, we will enforce the entire order of the Board and deny Hunter Douglas’ petition for review.